"git@developer.sourcefind.cn:tianlh/lightgbm-dcu.git" did not exist on "c591b77eb939222b2a9a98786a142aef787ffb61"
basic.py 163 KB
Newer Older
wxchan's avatar
wxchan committed
1
# coding: utf-8
2
"""Wrapper for C API of LightGBM."""
3
import abc
wxchan's avatar
wxchan committed
4
import ctypes
5
import json
wxchan's avatar
wxchan committed
6
import warnings
7
from collections import OrderedDict
8
from copy import deepcopy
9
10
from functools import wraps
from logging import Logger
11
12
13
from os import SEEK_END
from os.path import getsize
from pathlib import Path
14
from tempfile import NamedTemporaryFile
15
from typing import Any, Callable, Dict, Iterable, List, Optional, Set, Tuple, Union
wxchan's avatar
wxchan committed
16
17
18
19

import numpy as np
import scipy.sparse

20
from .compat import PANDAS_INSTALLED, concat, dt_DataTable, is_dtype_sparse, pd_DataFrame, pd_Series
wxchan's avatar
wxchan committed
21
22
from .libpath import find_lib_path

23
24
25
26
27
28
29
30
31
32
33
34
ZERO_THRESHOLD = 1e-35


def _get_sample_count(total_nrow: int, params: str):
    sample_cnt = ctypes.c_int(0)
    _safe_call(_LIB.LGBM_GetSampleCount(
        ctypes.c_int32(total_nrow),
        c_str(params),
        ctypes.byref(sample_cnt),
    ))
    return sample_cnt.value

wxchan's avatar
wxchan committed
35

36
class _DummyLogger:
37
    def info(self, msg: str) -> None:
38
39
        print(msg)

40
    def warning(self, msg: str) -> None:
41
42
43
        warnings.warn(msg, stacklevel=3)


44
_LOGGER: Union[_DummyLogger, Logger] = _DummyLogger()
45
46


47
def register_logger(logger: Logger) -> None:
48
49
50
51
52
53
54
55
56
57
58
59
60
    """Register custom logger.

    Parameters
    ----------
    logger : logging.Logger
        Custom logger.
    """
    if not isinstance(logger, Logger):
        raise TypeError("Logger should inherit logging.Logger class")
    global _LOGGER
    _LOGGER = logger


61
def _normalize_native_string(func: Callable[[str], None]) -> Callable[[str], None]:
62
    """Join log messages from native library which come by chunks."""
63
    msg_normalized: List[str] = []
64
65

    @wraps(func)
66
    def wrapper(msg: str) -> None:
67
68
69
70
71
72
73
74
75
76
77
        nonlocal msg_normalized
        if msg.strip() == '':
            msg = ''.join(msg_normalized)
            msg_normalized = []
            return func(msg)
        else:
            msg_normalized.append(msg)

    return wrapper


78
def _log_info(msg: str) -> None:
79
80
81
    _LOGGER.info(msg)


82
def _log_warning(msg: str) -> None:
83
84
85
86
    _LOGGER.warning(msg)


@_normalize_native_string
87
def _log_native(msg: str) -> None:
88
89
90
    _LOGGER.info(msg)


91
def _log_callback(msg: bytes) -> None:
92
    """Redirect logs from native library into Python."""
93
    _log_native(str(msg.decode('utf-8')))
94
95


wxchan's avatar
wxchan committed
96
def _load_lib():
97
    """Load LightGBM library."""
wxchan's avatar
wxchan committed
98
99
    lib_path = find_lib_path()
    if len(lib_path) == 0:
100
        return None
wxchan's avatar
wxchan committed
101
102
    lib = ctypes.cdll.LoadLibrary(lib_path[0])
    lib.LGBM_GetLastError.restype = ctypes.c_char_p
103
104
105
    callback = ctypes.CFUNCTYPE(None, ctypes.c_char_p)
    lib.callback = callback(_log_callback)
    if lib.LGBM_RegisterLogCallback(lib.callback) != 0:
106
        raise LightGBMError(lib.LGBM_GetLastError().decode('utf-8'))
wxchan's avatar
wxchan committed
107
108
    return lib

wxchan's avatar
wxchan committed
109

wxchan's avatar
wxchan committed
110
111
_LIB = _load_lib()

wxchan's avatar
wxchan committed
112

113
114
115
NUMERIC_TYPES = (int, float, bool)


116
def _safe_call(ret: int) -> None:
117
118
    """Check the return value from C API call.

wxchan's avatar
wxchan committed
119
120
121
    Parameters
    ----------
    ret : int
122
        The return value from C API calls.
wxchan's avatar
wxchan committed
123
124
    """
    if ret != 0:
125
        raise LightGBMError(_LIB.LGBM_GetLastError().decode('utf-8'))
wxchan's avatar
wxchan committed
126

wxchan's avatar
wxchan committed
127

wxchan's avatar
wxchan committed
128
def is_numeric(obj):
129
    """Check whether object is a number or not, include numpy number, etc."""
wxchan's avatar
wxchan committed
130
131
132
    try:
        float(obj)
        return True
wxchan's avatar
wxchan committed
133
134
135
    except (TypeError, ValueError):
        # TypeError: obj is not a string or a number
        # ValueError: invalid literal
wxchan's avatar
wxchan committed
136
137
        return False

wxchan's avatar
wxchan committed
138

wxchan's avatar
wxchan committed
139
def is_numpy_1d_array(data):
140
    """Check whether data is a numpy 1-D array."""
141
    return isinstance(data, np.ndarray) and len(data.shape) == 1
wxchan's avatar
wxchan committed
142

wxchan's avatar
wxchan committed
143

144
145
146
147
148
149
150
151
def is_numpy_column_array(data):
    """Check whether data is a column numpy array."""
    if not isinstance(data, np.ndarray):
        return False
    shape = data.shape
    return len(shape) == 2 and shape[1] == 1


152
153
def cast_numpy_array_to_dtype(array, dtype):
    """Cast numpy array to given dtype."""
154
155
156
157
158
    if array.dtype == dtype:
        return array
    return array.astype(dtype=dtype, copy=False)


wxchan's avatar
wxchan committed
159
def is_1d_list(data):
160
161
    """Check whether data is a 1-D list."""
    return isinstance(data, list) and (not data or is_numeric(data[0]))
wxchan's avatar
wxchan committed
162

wxchan's avatar
wxchan committed
163

164
165
166
167
168
169
170
171
172
173
def _is_1d_collection(data: Any) -> bool:
    """Check whether data is a 1-D collection."""
    return (
        is_numpy_1d_array(data)
        or is_numpy_column_array(data)
        or is_1d_list(data)
        or isinstance(data, pd_Series)
    )


174
def list_to_1d_numpy(data, dtype=np.float32, name='list'):
175
    """Convert data to numpy 1-D array."""
wxchan's avatar
wxchan committed
176
    if is_numpy_1d_array(data):
177
        return cast_numpy_array_to_dtype(data, dtype)
178
179
180
    elif is_numpy_column_array(data):
        _log_warning('Converting column-vector to 1d array')
        array = data.ravel()
181
        return cast_numpy_array_to_dtype(array, dtype)
wxchan's avatar
wxchan committed
182
183
    elif is_1d_list(data):
        return np.array(data, dtype=dtype, copy=False)
184
    elif isinstance(data, pd_Series):
185
186
        if _get_bad_pandas_dtypes([data.dtypes]):
            raise ValueError('Series.dtypes must be int, float or bool')
187
        return np.array(data, dtype=dtype, copy=False)  # SparseArray should be supported as well
wxchan's avatar
wxchan committed
188
    else:
189
190
        raise TypeError(f"Wrong type({type(data).__name__}) for {name}.\n"
                        "It should be list, numpy 1-D array or pandas Series")
wxchan's avatar
wxchan committed
191

wxchan's avatar
wxchan committed
192

193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
def _is_numpy_2d_array(data: Any) -> bool:
    """Check whether data is a numpy 2-D array."""
    return isinstance(data, np.ndarray) and len(data.shape) == 2 and data.shape[1] > 1


def _is_2d_list(data: Any) -> bool:
    """Check whether data is a 2-D list."""
    return isinstance(data, list) and len(data) > 0 and is_1d_list(data[0])


def _is_2d_collection(data: Any) -> bool:
    """Check whether data is a 2-D collection."""
    return (
        _is_numpy_2d_array(data)
        or _is_2d_list(data)
        or isinstance(data, pd_DataFrame)
    )


def _data_to_2d_numpy(data: Any, dtype: type = np.float32, name: str = 'list') -> np.ndarray:
    """Convert data to numpy 2-D array."""
    if _is_numpy_2d_array(data):
        return cast_numpy_array_to_dtype(data, dtype)
    if _is_2d_list(data):
        return np.array(data, dtype=dtype)
    if isinstance(data, pd_DataFrame):
        if _get_bad_pandas_dtypes(data.dtypes):
            raise ValueError('DataFrame.dtypes must be int, float or bool')
        return cast_numpy_array_to_dtype(data.values, dtype)
    raise TypeError(f"Wrong type({type(data).__name__}) for {name}.\n"
                    "It should be list of lists, numpy 2-D array or pandas DataFrame")


wxchan's avatar
wxchan committed
226
def cfloat32_array_to_numpy(cptr, length):
227
    """Convert a ctypes float pointer array to a numpy array."""
wxchan's avatar
wxchan committed
228
    if isinstance(cptr, ctypes.POINTER(ctypes.c_float)):
229
        return np.ctypeslib.as_array(cptr, shape=(length,)).copy()
wxchan's avatar
wxchan committed
230
    else:
231
        raise RuntimeError('Expected float pointer')
wxchan's avatar
wxchan committed
232

Guolin Ke's avatar
Guolin Ke committed
233

Guolin Ke's avatar
Guolin Ke committed
234
def cfloat64_array_to_numpy(cptr, length):
235
    """Convert a ctypes double pointer array to a numpy array."""
Guolin Ke's avatar
Guolin Ke committed
236
    if isinstance(cptr, ctypes.POINTER(ctypes.c_double)):
237
        return np.ctypeslib.as_array(cptr, shape=(length,)).copy()
Guolin Ke's avatar
Guolin Ke committed
238
239
240
    else:
        raise RuntimeError('Expected double pointer')

wxchan's avatar
wxchan committed
241

wxchan's avatar
wxchan committed
242
def cint32_array_to_numpy(cptr, length):
243
    """Convert a ctypes int pointer array to a numpy array."""
wxchan's avatar
wxchan committed
244
    if isinstance(cptr, ctypes.POINTER(ctypes.c_int32)):
245
        return np.ctypeslib.as_array(cptr, shape=(length,)).copy()
wxchan's avatar
wxchan committed
246
    else:
247
248
249
250
251
252
        raise RuntimeError('Expected int32 pointer')


def cint64_array_to_numpy(cptr, length):
    """Convert a ctypes int pointer array to a numpy array."""
    if isinstance(cptr, ctypes.POINTER(ctypes.c_int64)):
253
        return np.ctypeslib.as_array(cptr, shape=(length,)).copy()
254
255
    else:
        raise RuntimeError('Expected int64 pointer')
wxchan's avatar
wxchan committed
256

wxchan's avatar
wxchan committed
257

wxchan's avatar
wxchan committed
258
def c_str(string):
259
    """Convert a Python string to C string."""
wxchan's avatar
wxchan committed
260
261
    return ctypes.c_char_p(string.encode('utf-8'))

wxchan's avatar
wxchan committed
262

wxchan's avatar
wxchan committed
263
def c_array(ctype, values):
264
    """Convert a Python array to C array."""
wxchan's avatar
wxchan committed
265
266
    return (ctype * len(values))(*values)

wxchan's avatar
wxchan committed
267

268
269
270
271
272
273
274
275
276
277
def json_default_with_numpy(obj):
    """Convert numpy classes to JSON serializable objects."""
    if isinstance(obj, (np.integer, np.floating, np.bool_)):
        return obj.item()
    elif isinstance(obj, np.ndarray):
        return obj.tolist()
    else:
        return obj


wxchan's avatar
wxchan committed
278
def param_dict_to_str(data):
279
    """Convert Python dictionary to string, which is passed to C API."""
280
    if data is None or not data:
wxchan's avatar
wxchan committed
281
282
283
        return ""
    pairs = []
    for key, val in data.items():
284
        if isinstance(val, (list, tuple, set)) or is_numpy_1d_array(val):
285
286
            def to_string(x):
                if isinstance(x, list):
287
                    return f"[{','.join(map(str, x))}]"
288
289
                else:
                    return str(x)
290
            pairs.append(f"{key}={','.join(map(to_string, val))}")
291
        elif isinstance(val, (str, Path, NUMERIC_TYPES)) or is_numeric(val):
292
            pairs.append(f"{key}={val}")
293
        elif val is not None:
294
            raise TypeError(f'Unknown type of parameter:{key}, got:{type(val).__name__}')
wxchan's avatar
wxchan committed
295
    return ' '.join(pairs)
296

wxchan's avatar
wxchan committed
297

298
class _TempFile:
299
300
    """Proxy class to workaround errors on Windows."""

301
302
303
    def __enter__(self):
        with NamedTemporaryFile(prefix="lightgbm_tmp_", delete=True) as f:
            self.name = f.name
304
            self.path = Path(self.name)
305
        return self
wxchan's avatar
wxchan committed
306

307
    def __exit__(self, exc_type, exc_val, exc_tb):
308
309
        if self.path.is_file():
            self.path.unlink()
310

wxchan's avatar
wxchan committed
311

312
class LightGBMError(Exception):
313
314
    """Error thrown by LightGBM."""

315
316
317
    pass


318
319
320
321
322
323
324
325
# DeprecationWarning is not shown by default, so let's create our own with higher level
class LGBMDeprecationWarning(UserWarning):
    """Custom deprecation warning."""

    pass


class _ConfigAliases:
326
327
328
    aliases = {"bin_construct_sample_cnt": {"bin_construct_sample_cnt",
                                            "subsample_for_bin"},
               "boosting": {"boosting",
329
330
331
332
333
                            "boosting_type",
                            "boost"},
               "categorical_feature": {"categorical_feature",
                                       "cat_feature",
                                       "categorical_column",
334
335
                                       "cat_column",
                                       "categorical_features"},
336
337
               "data_random_seed": {"data_random_seed",
                                    "data_seed"},
338
339
340
341
               "early_stopping_round": {"early_stopping_round",
                                        "early_stopping_rounds",
                                        "early_stopping",
                                        "n_iter_no_change"},
342
343
344
               "enable_bundle": {"enable_bundle",
                                 "is_enable_bundle",
                                 "bundle"},
345
346
347
348
349
               "eval_at": {"eval_at",
                           "ndcg_eval_at",
                           "ndcg_at",
                           "map_eval_at",
                           "map_at"},
350
351
352
353
354
355
               "group_column": {"group_column",
                                "group",
                                "group_id",
                                "query_column",
                                "query",
                                "query_id"},
356
357
               "header": {"header",
                          "has_header"},
358
359
360
361
362
363
364
365
366
               "ignore_column": {"ignore_column",
                                 "ignore_feature",
                                 "blacklist"},
               "is_enable_sparse": {"is_enable_sparse",
                                    "is_sparse",
                                    "enable_sparse",
                                    "sparse"},
               "label_column": {"label_column",
                                "label"},
Nikita Titov's avatar
Nikita Titov committed
367
368
               "linear_tree": {"linear_tree",
                               "linear_trees"},
369
370
371
               "local_listen_port": {"local_listen_port",
                                     "local_port",
                                     "port"},
372
373
374
               "machines": {"machines",
                            "workers",
                            "nodes"},
375
376
               "max_bin": {"max_bin",
                           "max_bins"},
377
378
379
380
381
382
383
384
385
386
387
388
389
               "metric": {"metric",
                          "metrics",
                          "metric_types"},
               "num_class": {"num_class",
                             "num_classes"},
               "num_iterations": {"num_iterations",
                                  "num_iteration",
                                  "n_iter",
                                  "num_tree",
                                  "num_trees",
                                  "num_round",
                                  "num_rounds",
                                  "num_boost_round",
390
391
                                  "n_estimators",
                                  "max_iter"},
392
393
               "num_machines": {"num_machines",
                                "num_machine"},
394
395
396
397
398
               "num_threads": {"num_threads",
                               "num_thread",
                               "nthread",
                               "nthreads",
                               "n_jobs"},
399
400
401
               "objective": {"objective",
                             "objective_type",
                             "app",
402
403
                             "application",
                             "loss"},
404
405
               "pre_partition": {"pre_partition",
                                 "is_pre_partition"},
406
407
408
409
               "tree_learner": {"tree_learner",
                                "tree",
                                "tree_type",
                                "tree_learner_type"},
410
411
412
               "two_round": {"two_round",
                             "two_round_loading",
                             "use_two_round_loading"},
413
               "verbosity": {"verbosity",
414
415
416
                             "verbose"},
               "weight_column": {"weight_column",
                                 "weight"}}
417
418
419
420
421

    @classmethod
    def get(cls, *args):
        ret = set()
        for i in args:
422
            ret |= cls.aliases.get(i, {i})
423
424
425
        return ret


426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
def _choose_param_value(main_param_name: str, params: Dict[str, Any], default_value: Any) -> Dict[str, Any]:
    """Get a single parameter value, accounting for aliases.

    Parameters
    ----------
    main_param_name : str
        Name of the main parameter to get a value for. One of the keys of ``_ConfigAliases``.
    params : dict
        Dictionary of LightGBM parameters.
    default_value : Any
        Default value to use for the parameter, if none is found in ``params``.

    Returns
    -------
    params : dict
        A ``params`` dict with exactly one value for ``main_param_name``, and all aliases ``main_param_name`` removed.
        If both ``main_param_name`` and one or more aliases for it are found, the value of ``main_param_name`` will be preferred.
    """
    # avoid side effects on passed-in parameters
    params = deepcopy(params)

    # find a value, and remove other aliases with .pop()
    # prefer the value of 'main_param_name' if it exists, otherwise search the aliases
    found_value = None
    if main_param_name in params.keys():
        found_value = params[main_param_name]

    for param in _ConfigAliases.get(main_param_name):
        val = params.pop(param, None)
        if found_value is None and val is not None:
            found_value = val

    if found_value is not None:
        params[main_param_name] = found_value
    else:
        params[main_param_name] = default_value

    return params


466
467
MAX_INT32 = (1 << 31) - 1

468
"""Macro definition of data type in C API of LightGBM"""
wxchan's avatar
wxchan committed
469
470
471
472
C_API_DTYPE_FLOAT32 = 0
C_API_DTYPE_FLOAT64 = 1
C_API_DTYPE_INT32 = 2
C_API_DTYPE_INT64 = 3
Guolin Ke's avatar
Guolin Ke committed
473

474
"""Matrix is row major in Python"""
wxchan's avatar
wxchan committed
475
476
C_API_IS_ROW_MAJOR = 1

477
"""Macro definition of prediction type in C API of LightGBM"""
wxchan's avatar
wxchan committed
478
479
480
C_API_PREDICT_NORMAL = 0
C_API_PREDICT_RAW_SCORE = 1
C_API_PREDICT_LEAF_INDEX = 2
481
C_API_PREDICT_CONTRIB = 3
wxchan's avatar
wxchan committed
482

483
484
485
486
"""Macro definition of sparse matrix type"""
C_API_MATRIX_TYPE_CSR = 0
C_API_MATRIX_TYPE_CSC = 1

487
488
489
490
"""Macro definition of feature importance type"""
C_API_FEATURE_IMPORTANCE_SPLIT = 0
C_API_FEATURE_IMPORTANCE_GAIN = 1

491
"""Data type of data field"""
wxchan's avatar
wxchan committed
492
493
FIELD_TYPE_MAPPER = {"label": C_API_DTYPE_FLOAT32,
                     "weight": C_API_DTYPE_FLOAT32,
Guolin Ke's avatar
Guolin Ke committed
494
                     "init_score": C_API_DTYPE_FLOAT64,
495
                     "group": C_API_DTYPE_INT32}
wxchan's avatar
wxchan committed
496

497
498
499
500
"""String name to int feature importance type mapper"""
FEATURE_IMPORTANCE_TYPE_MAPPER = {"split": C_API_FEATURE_IMPORTANCE_SPLIT,
                                  "gain": C_API_FEATURE_IMPORTANCE_GAIN}

wxchan's avatar
wxchan committed
501

502
def convert_from_sliced_object(data):
503
    """Fix the memory of multi-dimensional sliced object."""
504
    if isinstance(data, np.ndarray) and isinstance(data.base, np.ndarray):
505
        if not data.flags.c_contiguous:
506
507
            _log_warning("Usage of np.ndarray subset (sliced data) is not recommended "
                         "due to it will double the peak memory cost in LightGBM.")
508
509
510
511
            return np.copy(data)
    return data


wxchan's avatar
wxchan committed
512
def c_float_array(data):
513
    """Get pointer of float numpy array / list."""
wxchan's avatar
wxchan committed
514
515
516
    if is_1d_list(data):
        data = np.array(data, copy=False)
    if is_numpy_1d_array(data):
517
518
        data = convert_from_sliced_object(data)
        assert data.flags.c_contiguous
wxchan's avatar
wxchan committed
519
520
521
522
523
524
525
        if data.dtype == np.float32:
            ptr_data = data.ctypes.data_as(ctypes.POINTER(ctypes.c_float))
            type_data = C_API_DTYPE_FLOAT32
        elif data.dtype == np.float64:
            ptr_data = data.ctypes.data_as(ctypes.POINTER(ctypes.c_double))
            type_data = C_API_DTYPE_FLOAT64
        else:
526
            raise TypeError(f"Expected np.float32 or np.float64, met type({data.dtype})")
wxchan's avatar
wxchan committed
527
    else:
528
        raise TypeError(f"Unknown type({type(data).__name__})")
529
    return (ptr_data, type_data, data)  # return `data` to avoid the temporary copy is freed
wxchan's avatar
wxchan committed
530

wxchan's avatar
wxchan committed
531

wxchan's avatar
wxchan committed
532
def c_int_array(data):
533
    """Get pointer of int numpy array / list."""
wxchan's avatar
wxchan committed
534
535
536
    if is_1d_list(data):
        data = np.array(data, copy=False)
    if is_numpy_1d_array(data):
537
538
        data = convert_from_sliced_object(data)
        assert data.flags.c_contiguous
wxchan's avatar
wxchan committed
539
540
541
542
543
544
545
        if data.dtype == np.int32:
            ptr_data = data.ctypes.data_as(ctypes.POINTER(ctypes.c_int32))
            type_data = C_API_DTYPE_INT32
        elif data.dtype == np.int64:
            ptr_data = data.ctypes.data_as(ctypes.POINTER(ctypes.c_int64))
            type_data = C_API_DTYPE_INT64
        else:
546
            raise TypeError(f"Expected np.int32 or np.int64, met type({data.dtype})")
wxchan's avatar
wxchan committed
547
    else:
548
        raise TypeError(f"Unknown type({type(data).__name__})")
549
    return (ptr_data, type_data, data)  # return `data` to avoid the temporary copy is freed
wxchan's avatar
wxchan committed
550

wxchan's avatar
wxchan committed
551

552
553
554
555
556
557
558
559
560
561
562
def _get_bad_pandas_dtypes(dtypes):
    pandas_dtype_mapper = {'int8': 'int', 'int16': 'int', 'int32': 'int',
                           'int64': 'int', 'uint8': 'int', 'uint16': 'int',
                           'uint32': 'int', 'uint64': 'int', 'bool': 'int',
                           'float16': 'float', 'float32': 'float', 'float64': 'float'}
    bad_indices = [i for i, dtype in enumerate(dtypes) if (dtype.name not in pandas_dtype_mapper
                                                           and (not is_dtype_sparse(dtype)
                                                                or dtype.subtype.name not in pandas_dtype_mapper))]
    return bad_indices


563
def _data_from_pandas(data, feature_name, categorical_feature, pandas_categorical):
564
    if isinstance(data, pd_DataFrame):
565
566
        if len(data.shape) != 2 or data.shape[0] < 1:
            raise ValueError('Input data must be 2 dimensional and non empty.')
567
568
        if feature_name == 'auto' or feature_name is None:
            data = data.rename(columns=str)
569
570
        cat_cols = list(data.select_dtypes(include=['category']).columns)
        cat_cols_not_ordered = [col for col in cat_cols if not data[col].cat.ordered]
571
572
573
574
575
        if pandas_categorical is None:  # train dataset
            pandas_categorical = [list(data[col].cat.categories) for col in cat_cols]
        else:
            if len(cat_cols) != len(pandas_categorical):
                raise ValueError('train and valid dataset categorical_feature do not match.')
576
            for col, category in zip(cat_cols, pandas_categorical):
577
578
                if list(data[col].cat.categories) != list(category):
                    data[col] = data[col].cat.set_categories(category)
579
        if len(cat_cols):  # cat_cols is list
580
            data = data.copy()  # not alter origin DataFrame
581
            data[cat_cols] = data[cat_cols].apply(lambda x: x.cat.codes).replace({-1: np.nan})
582
583
584
        if categorical_feature is not None:
            if feature_name is None:
                feature_name = list(data.columns)
585
            if categorical_feature == 'auto':  # use cat cols from DataFrame
586
                categorical_feature = cat_cols_not_ordered
587
588
            else:  # use cat cols specified by user
                categorical_feature = list(categorical_feature)
589
590
        if feature_name == 'auto':
            feature_name = list(data.columns)
591
592
        bad_indices = _get_bad_pandas_dtypes(data.dtypes)
        if bad_indices:
593
            bad_index_cols_str = ', '.join(data.columns[bad_indices])
594
            raise ValueError("DataFrame.dtypes for data must be int, float or bool.\n"
595
                             "Did not expect the data types in the following fields: "
596
                             f"{bad_index_cols_str}")
597
598
599
        data = data.values
        if data.dtype != np.float32 and data.dtype != np.float64:
            data = data.astype(np.float32)
600
601
602
603
604
605
    else:
        if feature_name == 'auto':
            feature_name = None
        if categorical_feature == 'auto':
            categorical_feature = None
    return data, feature_name, categorical_feature, pandas_categorical
606
607
608


def _label_from_pandas(label):
609
    if isinstance(label, pd_DataFrame):
610
611
        if len(label.columns) > 1:
            raise ValueError('DataFrame for label cannot have multiple columns')
612
        if _get_bad_pandas_dtypes(label.dtypes):
613
            raise ValueError('DataFrame.dtypes for label must be int, float or bool')
614
        label = np.ravel(label.values.astype(np.float32, copy=False))
615
616
617
    return label


618
def _dump_pandas_categorical(pandas_categorical, file_name=None):
619
620
    categorical_json = json.dumps(pandas_categorical, default=json_default_with_numpy)
    pandas_str = f'\npandas_categorical:{categorical_json}\n'
621
622
623
624
625
626
627
    if file_name is not None:
        with open(file_name, 'a') as f:
            f.write(pandas_str)
    return pandas_str


def _load_pandas_categorical(file_name=None, model_str=None):
628
629
    pandas_key = 'pandas_categorical:'
    offset = -len(pandas_key)
630
    if file_name is not None:
631
        max_offset = -getsize(file_name)
632
633
634
635
        with open(file_name, 'rb') as f:
            while True:
                if offset < max_offset:
                    offset = max_offset
636
                f.seek(offset, SEEK_END)
637
638
639
640
                lines = f.readlines()
                if len(lines) >= 2:
                    break
                offset *= 2
641
        last_line = lines[-1].decode('utf-8').strip()
642
        if not last_line.startswith(pandas_key):
643
            last_line = lines[-2].decode('utf-8').strip()
644
    elif model_str is not None:
645
646
647
648
649
650
        idx = model_str.rfind('\n', 0, offset)
        last_line = model_str[idx:].strip()
    if last_line.startswith(pandas_key):
        return json.loads(last_line[len(pandas_key):])
    else:
        return None
651
652


653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
class Sequence(abc.ABC):
    """
    Generic data access interface.

    Object should support the following operations:

    .. code-block::

        # Get total row number.
        >>> len(seq)
        # Random access by row index. Used for data sampling.
        >>> seq[10]
        # Range data access. Used to read data in batch when constructing Dataset.
        >>> seq[0:100]
        # Optionally specify batch_size to control range data read size.
        >>> seq.batch_size

    - With random access, **data sampling does not need to go through all data**.
    - With range data access, there's **no need to read all data into memory thus reduce memory usage**.

673
674
    .. versionadded:: 3.3.0

675
676
677
678
679
680
681
682
683
    Attributes
    ----------
    batch_size : int
        Default size of a batch.
    """

    batch_size = 4096  # Defaults to read 4K rows in each batch.

    @abc.abstractmethod
684
    def __getitem__(self, idx: Union[int, slice, List[int]]) -> np.ndarray:
685
686
687
688
689
690
691
        """Return data for given row index.

        A basic implementation should look like this:

        .. code-block:: python

            if isinstance(idx, numbers.Integral):
692
                return self._get_one_line(idx)
693
            elif isinstance(idx, slice):
694
695
                return np.stack([self._get_one_line(i) for i in range(idx.start, idx.stop)])
            elif isinstance(idx, list):
696
                # Only required if using ``Dataset.subset()``.
697
                return np.array([self._get_one_line(i) for i in idx])
698
            else:
699
                raise TypeError(f"Sequence index must be integer, slice or list, got {type(idx).__name__}")
700
701
702

        Parameters
        ----------
703
        idx : int, slice[int], list[int]
704
705
706
707
708
            Item index.

        Returns
        -------
        result : numpy 1-D array, numpy 2-D array
709
            1-D array if idx is int, 2-D array if idx is slice or list.
710
711
712
713
714
715
716
717
718
        """
        raise NotImplementedError("Sub-classes of lightgbm.Sequence must implement __getitem__()")

    @abc.abstractmethod
    def __len__(self) -> int:
        """Return row count of this sequence."""
        raise NotImplementedError("Sub-classes of lightgbm.Sequence must implement __len__()")


719
class _InnerPredictor:
720
721
722
723
724
    """_InnerPredictor of LightGBM.

    Not exposed to user.
    Used only for prediction, usually used for continued training.

Nikita Titov's avatar
Nikita Titov committed
725
726
727
    .. note::

        Can be converted from Booster, but cannot be converted to Booster.
Guolin Ke's avatar
Guolin Ke committed
728
    """
729

730
    def __init__(self, model_file=None, booster_handle=None, pred_parameter=None):
731
        """Initialize the _InnerPredictor.
wxchan's avatar
wxchan committed
732
733
734

        Parameters
        ----------
735
        model_file : str, pathlib.Path or None, optional (default=None)
wxchan's avatar
wxchan committed
736
            Path to the model file.
737
738
739
        booster_handle : object or None, optional (default=None)
            Handle of Booster.
        pred_parameter: dict or None, optional (default=None)
740
            Other parameters for the prediction.
wxchan's avatar
wxchan committed
741
742
743
744
745
        """
        self.handle = ctypes.c_void_p()
        self.__is_manage_handle = True
        if model_file is not None:
            """Prediction task"""
Guolin Ke's avatar
Guolin Ke committed
746
            out_num_iterations = ctypes.c_int(0)
wxchan's avatar
wxchan committed
747
            _safe_call(_LIB.LGBM_BoosterCreateFromModelfile(
748
                c_str(str(model_file)),
wxchan's avatar
wxchan committed
749
750
                ctypes.byref(out_num_iterations),
                ctypes.byref(self.handle)))
Guolin Ke's avatar
Guolin Ke committed
751
            out_num_class = ctypes.c_int(0)
wxchan's avatar
wxchan committed
752
753
754
755
            _safe_call(_LIB.LGBM_BoosterGetNumClasses(
                self.handle,
                ctypes.byref(out_num_class)))
            self.num_class = out_num_class.value
756
            self.num_total_iteration = out_num_iterations.value
757
            self.pandas_categorical = _load_pandas_categorical(file_name=model_file)
wxchan's avatar
wxchan committed
758
        elif booster_handle is not None:
Guolin Ke's avatar
Guolin Ke committed
759
            self.__is_manage_handle = False
wxchan's avatar
wxchan committed
760
            self.handle = booster_handle
Guolin Ke's avatar
Guolin Ke committed
761
            out_num_class = ctypes.c_int(0)
wxchan's avatar
wxchan committed
762
763
764
765
            _safe_call(_LIB.LGBM_BoosterGetNumClasses(
                self.handle,
                ctypes.byref(out_num_class)))
            self.num_class = out_num_class.value
766
            self.num_total_iteration = self.current_iteration()
767
            self.pandas_categorical = None
wxchan's avatar
wxchan committed
768
        else:
769
            raise TypeError('Need model_file or booster_handle to create a predictor')
wxchan's avatar
wxchan committed
770

771
772
        pred_parameter = {} if pred_parameter is None else pred_parameter
        self.pred_parameter = param_dict_to_str(pred_parameter)
cbecker's avatar
cbecker committed
773

wxchan's avatar
wxchan committed
774
    def __del__(self):
775
776
777
778
779
        try:
            if self.__is_manage_handle:
                _safe_call(_LIB.LGBM_BoosterFree(self.handle))
        except AttributeError:
            pass
wxchan's avatar
wxchan committed
780

781
782
783
784
785
    def __getstate__(self):
        this = self.__dict__.copy()
        this.pop('handle', None)
        return this

786
    def predict(self, data, start_iteration=0, num_iteration=-1,
787
                raw_score=False, pred_leaf=False, pred_contrib=False, data_has_header=False,
wxchan's avatar
wxchan committed
788
                is_reshape=True):
789
        """Predict logic.
wxchan's avatar
wxchan committed
790
791
792

        Parameters
        ----------
793
        data : str, pathlib.Path, numpy array, pandas DataFrame, H2O DataTable's Frame or scipy.sparse
794
            Data source for prediction.
795
            If str or pathlib.Path, it represents the path to a text file (CSV, TSV, or LibSVM).
796
797
        start_iteration : int, optional (default=0)
            Start index of the iteration to predict.
798
799
800
801
802
803
804
805
806
807
808
809
810
        num_iteration : int, optional (default=-1)
            Iteration used for prediction.
        raw_score : bool, optional (default=False)
            Whether to predict raw scores.
        pred_leaf : bool, optional (default=False)
            Whether to predict leaf index.
        pred_contrib : bool, optional (default=False)
            Whether to predict feature contributions.
        data_has_header : bool, optional (default=False)
            Whether data has header.
            Used only for txt data.
        is_reshape : bool, optional (default=True)
            Whether to reshape to (nrow, ncol).
wxchan's avatar
wxchan committed
811
812
813

        Returns
        -------
814
        result : numpy array, scipy.sparse or list of scipy.sparse
815
            Prediction result.
816
            Can be sparse or a list of sparse objects (each element represents predictions for one class) for feature contributions (when ``pred_contrib=True``).
wxchan's avatar
wxchan committed
817
        """
wxchan's avatar
wxchan committed
818
        if isinstance(data, Dataset):
819
            raise TypeError("Cannot use Dataset instance for prediction, please use raw data instead")
820
        data = _data_from_pandas(data, None, None, self.pandas_categorical)[0]
wxchan's avatar
wxchan committed
821
822
823
824
825
        predict_type = C_API_PREDICT_NORMAL
        if raw_score:
            predict_type = C_API_PREDICT_RAW_SCORE
        if pred_leaf:
            predict_type = C_API_PREDICT_LEAF_INDEX
826
827
        if pred_contrib:
            predict_type = C_API_PREDICT_CONTRIB
wxchan's avatar
wxchan committed
828
        int_data_has_header = 1 if data_has_header else 0
cbecker's avatar
cbecker committed
829

830
        if isinstance(data, (str, Path)):
831
            with _TempFile() as f:
wxchan's avatar
wxchan committed
832
833
                _safe_call(_LIB.LGBM_BoosterPredictForFile(
                    self.handle,
834
                    c_str(str(data)),
Guolin Ke's avatar
Guolin Ke committed
835
836
                    ctypes.c_int(int_data_has_header),
                    ctypes.c_int(predict_type),
837
                    ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
838
                    ctypes.c_int(num_iteration),
839
                    c_str(self.pred_parameter),
wxchan's avatar
wxchan committed
840
                    c_str(f.name)))
841
842
                preds = np.loadtxt(f.name, dtype=np.float64)
                nrow = preds.shape[0]
wxchan's avatar
wxchan committed
843
        elif isinstance(data, scipy.sparse.csr_matrix):
844
            preds, nrow = self.__pred_for_csr(data, start_iteration, num_iteration, predict_type)
Guolin Ke's avatar
Guolin Ke committed
845
        elif isinstance(data, scipy.sparse.csc_matrix):
846
            preds, nrow = self.__pred_for_csc(data, start_iteration, num_iteration, predict_type)
wxchan's avatar
wxchan committed
847
        elif isinstance(data, np.ndarray):
848
            preds, nrow = self.__pred_for_np2d(data, start_iteration, num_iteration, predict_type)
849
850
851
        elif isinstance(data, list):
            try:
                data = np.array(data)
852
            except BaseException:
853
                raise ValueError('Cannot convert data list to numpy array.')
854
            preds, nrow = self.__pred_for_np2d(data, start_iteration, num_iteration, predict_type)
855
        elif isinstance(data, dt_DataTable):
856
            preds, nrow = self.__pred_for_np2d(data.to_numpy(), start_iteration, num_iteration, predict_type)
wxchan's avatar
wxchan committed
857
858
        else:
            try:
859
                _log_warning('Converting data to scipy sparse matrix.')
wxchan's avatar
wxchan committed
860
                csr = scipy.sparse.csr_matrix(data)
861
            except BaseException:
862
                raise TypeError(f'Cannot predict data for type {type(data).__name__}')
863
            preds, nrow = self.__pred_for_csr(csr, start_iteration, num_iteration, predict_type)
wxchan's avatar
wxchan committed
864
865
        if pred_leaf:
            preds = preds.astype(np.int32)
866
867
        is_sparse = scipy.sparse.issparse(preds) or isinstance(preds, list)
        if is_reshape and not is_sparse and preds.size != nrow:
wxchan's avatar
wxchan committed
868
            if preds.size % nrow == 0:
869
                preds = preds.reshape(nrow, -1)
wxchan's avatar
wxchan committed
870
            else:
871
                raise ValueError(f'Length of predict result ({preds.size}) cannot be divide nrow ({nrow})')
wxchan's avatar
wxchan committed
872
873
        return preds

874
    def __get_num_preds(self, start_iteration, num_iteration, nrow, predict_type):
875
        """Get size of prediction result."""
876
        if nrow > MAX_INT32:
877
            raise LightGBMError('LightGBM cannot perform prediction for data '
878
                                f'with number of rows greater than MAX_INT32 ({MAX_INT32}).\n'
879
                                'You can split your data into chunks '
880
                                'and then concatenate predictions for them')
Guolin Ke's avatar
Guolin Ke committed
881
882
883
        n_preds = ctypes.c_int64(0)
        _safe_call(_LIB.LGBM_BoosterCalcNumPredict(
            self.handle,
Guolin Ke's avatar
Guolin Ke committed
884
885
            ctypes.c_int(nrow),
            ctypes.c_int(predict_type),
886
            ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
887
            ctypes.c_int(num_iteration),
Guolin Ke's avatar
Guolin Ke committed
888
889
            ctypes.byref(n_preds)))
        return n_preds.value
wxchan's avatar
wxchan committed
890

891
    def __pred_for_np2d(self, mat, start_iteration, num_iteration, predict_type):
892
        """Predict for a 2-D numpy matrix."""
wxchan's avatar
wxchan committed
893
        if len(mat.shape) != 2:
894
            raise ValueError('Input numpy.ndarray or list must be 2 dimensional')
wxchan's avatar
wxchan committed
895

896
        def inner_predict(mat, start_iteration, num_iteration, predict_type, preds=None):
897
898
            if mat.dtype == np.float32 or mat.dtype == np.float64:
                data = np.array(mat.reshape(mat.size), dtype=mat.dtype, copy=False)
899
            else:  # change non-float data to float data, need to copy
900
901
                data = np.array(mat.reshape(mat.size), dtype=np.float32)
            ptr_data, type_ptr_data, _ = c_float_array(data)
902
            n_preds = self.__get_num_preds(start_iteration, num_iteration, mat.shape[0], predict_type)
903
            if preds is None:
904
                preds = np.empty(n_preds, dtype=np.float64)
905
906
907
908
909
910
911
            elif len(preds.shape) != 1 or len(preds) != n_preds:
                raise ValueError("Wrong length of pre-allocated predict array")
            out_num_preds = ctypes.c_int64(0)
            _safe_call(_LIB.LGBM_BoosterPredictForMat(
                self.handle,
                ptr_data,
                ctypes.c_int(type_ptr_data),
912
913
                ctypes.c_int32(mat.shape[0]),
                ctypes.c_int32(mat.shape[1]),
914
915
                ctypes.c_int(C_API_IS_ROW_MAJOR),
                ctypes.c_int(predict_type),
916
                ctypes.c_int(start_iteration),
917
918
919
920
921
922
923
924
925
926
927
928
                ctypes.c_int(num_iteration),
                c_str(self.pred_parameter),
                ctypes.byref(out_num_preds),
                preds.ctypes.data_as(ctypes.POINTER(ctypes.c_double))))
            if n_preds != out_num_preds.value:
                raise ValueError("Wrong length for predict results")
            return preds, mat.shape[0]

        nrow = mat.shape[0]
        if nrow > MAX_INT32:
            sections = np.arange(start=MAX_INT32, stop=nrow, step=MAX_INT32)
            # __get_num_preds() cannot work with nrow > MAX_INT32, so calculate overall number of predictions piecemeal
929
            n_preds = [self.__get_num_preds(start_iteration, num_iteration, i, predict_type) for i in np.diff([0] + list(sections) + [nrow])]
930
            n_preds_sections = np.array([0] + n_preds, dtype=np.intp).cumsum()
931
            preds = np.empty(sum(n_preds), dtype=np.float64)
932
933
            for chunk, (start_idx_pred, end_idx_pred) in zip(np.array_split(mat, sections),
                                                             zip(n_preds_sections, n_preds_sections[1:])):
934
                # avoid memory consumption by arrays concatenation operations
935
                inner_predict(chunk, start_iteration, num_iteration, predict_type, preds[start_idx_pred:end_idx_pred])
936
            return preds, nrow
wxchan's avatar
wxchan committed
937
        else:
938
            return inner_predict(mat, start_iteration, num_iteration, predict_type)
wxchan's avatar
wxchan committed
939

940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
    def __create_sparse_native(self, cs, out_shape, out_ptr_indptr, out_ptr_indices, out_ptr_data,
                               indptr_type, data_type, is_csr=True):
        # create numpy array from output arrays
        data_indices_len = out_shape[0]
        indptr_len = out_shape[1]
        if indptr_type == C_API_DTYPE_INT32:
            out_indptr = cint32_array_to_numpy(out_ptr_indptr, indptr_len)
        elif indptr_type == C_API_DTYPE_INT64:
            out_indptr = cint64_array_to_numpy(out_ptr_indptr, indptr_len)
        else:
            raise TypeError("Expected int32 or int64 type for indptr")
        if data_type == C_API_DTYPE_FLOAT32:
            out_data = cfloat32_array_to_numpy(out_ptr_data, data_indices_len)
        elif data_type == C_API_DTYPE_FLOAT64:
            out_data = cfloat64_array_to_numpy(out_ptr_data, data_indices_len)
        else:
            raise TypeError("Expected float32 or float64 type for data")
        out_indices = cint32_array_to_numpy(out_ptr_indices, data_indices_len)
        # break up indptr based on number of rows (note more than one matrix in multiclass case)
        per_class_indptr_shape = cs.indptr.shape[0]
        # for CSC there is extra column added
        if not is_csr:
            per_class_indptr_shape += 1
        out_indptr_arrays = np.split(out_indptr, out_indptr.shape[0] / per_class_indptr_shape)
        # reformat output into a csr or csc matrix or list of csr or csc matrices
        cs_output_matrices = []
        offset = 0
        for cs_indptr in out_indptr_arrays:
            matrix_indptr_len = cs_indptr[cs_indptr.shape[0] - 1]
            cs_indices = out_indices[offset + cs_indptr[0]:offset + matrix_indptr_len]
            cs_data = out_data[offset + cs_indptr[0]:offset + matrix_indptr_len]
            offset += matrix_indptr_len
            # same shape as input csr or csc matrix except extra column for expected value
            cs_shape = [cs.shape[0], cs.shape[1] + 1]
            # note: make sure we copy data as it will be deallocated next
            if is_csr:
                cs_output_matrices.append(scipy.sparse.csr_matrix((cs_data, cs_indices, cs_indptr), cs_shape))
            else:
                cs_output_matrices.append(scipy.sparse.csc_matrix((cs_data, cs_indices, cs_indptr), cs_shape))
        # free the temporary native indptr, indices, and data
        _safe_call(_LIB.LGBM_BoosterFreePredictSparse(out_ptr_indptr, out_ptr_indices, out_ptr_data,
                                                      ctypes.c_int(indptr_type), ctypes.c_int(data_type)))
        if len(cs_output_matrices) == 1:
            return cs_output_matrices[0]
        return cs_output_matrices

986
    def __pred_for_csr(self, csr, start_iteration, num_iteration, predict_type):
987
        """Predict for a CSR data."""
988
        def inner_predict(csr, start_iteration, num_iteration, predict_type, preds=None):
989
            nrow = len(csr.indptr) - 1
990
            n_preds = self.__get_num_preds(start_iteration, num_iteration, nrow, predict_type)
991
            if preds is None:
992
                preds = np.empty(n_preds, dtype=np.float64)
993
994
995
996
997
998
999
            elif len(preds.shape) != 1 or len(preds) != n_preds:
                raise ValueError("Wrong length of pre-allocated predict array")
            out_num_preds = ctypes.c_int64(0)

            ptr_indptr, type_ptr_indptr, __ = c_int_array(csr.indptr)
            ptr_data, type_ptr_data, _ = c_float_array(csr.data)

1000
            assert csr.shape[1] <= MAX_INT32
1001
            csr_indices = csr.indices.astype(np.int32, copy=False)
1002

1003
1004
1005
            _safe_call(_LIB.LGBM_BoosterPredictForCSR(
                self.handle,
                ptr_indptr,
1006
                ctypes.c_int(type_ptr_indptr),
1007
                csr_indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
1008
1009
1010
1011
1012
1013
                ptr_data,
                ctypes.c_int(type_ptr_data),
                ctypes.c_int64(len(csr.indptr)),
                ctypes.c_int64(len(csr.data)),
                ctypes.c_int64(csr.shape[1]),
                ctypes.c_int(predict_type),
1014
                ctypes.c_int(start_iteration),
1015
1016
1017
1018
1019
1020
1021
                ctypes.c_int(num_iteration),
                c_str(self.pred_parameter),
                ctypes.byref(out_num_preds),
                preds.ctypes.data_as(ctypes.POINTER(ctypes.c_double))))
            if n_preds != out_num_preds.value:
                raise ValueError("Wrong length for predict results")
            return preds, nrow
wxchan's avatar
wxchan committed
1022

1023
        def inner_predict_sparse(csr, start_iteration, num_iteration, predict_type):
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
            ptr_indptr, type_ptr_indptr, __ = c_int_array(csr.indptr)
            ptr_data, type_ptr_data, _ = c_float_array(csr.data)
            csr_indices = csr.indices.astype(np.int32, copy=False)
            matrix_type = C_API_MATRIX_TYPE_CSR
            if type_ptr_indptr == C_API_DTYPE_INT32:
                out_ptr_indptr = ctypes.POINTER(ctypes.c_int32)()
            else:
                out_ptr_indptr = ctypes.POINTER(ctypes.c_int64)()
            out_ptr_indices = ctypes.POINTER(ctypes.c_int32)()
            if type_ptr_data == C_API_DTYPE_FLOAT32:
                out_ptr_data = ctypes.POINTER(ctypes.c_float)()
            else:
                out_ptr_data = ctypes.POINTER(ctypes.c_double)()
1037
            out_shape = np.empty(2, dtype=np.int64)
1038
1039
1040
            _safe_call(_LIB.LGBM_BoosterPredictSparseOutput(
                self.handle,
                ptr_indptr,
1041
                ctypes.c_int(type_ptr_indptr),
1042
1043
1044
1045
1046
1047
1048
                csr_indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
                ptr_data,
                ctypes.c_int(type_ptr_data),
                ctypes.c_int64(len(csr.indptr)),
                ctypes.c_int64(len(csr.data)),
                ctypes.c_int64(csr.shape[1]),
                ctypes.c_int(predict_type),
1049
                ctypes.c_int(start_iteration),
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
                ctypes.c_int(num_iteration),
                c_str(self.pred_parameter),
                ctypes.c_int(matrix_type),
                out_shape.ctypes.data_as(ctypes.POINTER(ctypes.c_int64)),
                ctypes.byref(out_ptr_indptr),
                ctypes.byref(out_ptr_indices),
                ctypes.byref(out_ptr_data)))
            matrices = self.__create_sparse_native(csr, out_shape, out_ptr_indptr, out_ptr_indices, out_ptr_data,
                                                   type_ptr_indptr, type_ptr_data, is_csr=True)
            nrow = len(csr.indptr) - 1
            return matrices, nrow

        if predict_type == C_API_PREDICT_CONTRIB:
1063
            return inner_predict_sparse(csr, start_iteration, num_iteration, predict_type)
1064
1065
1066
1067
        nrow = len(csr.indptr) - 1
        if nrow > MAX_INT32:
            sections = [0] + list(np.arange(start=MAX_INT32, stop=nrow, step=MAX_INT32)) + [nrow]
            # __get_num_preds() cannot work with nrow > MAX_INT32, so calculate overall number of predictions piecemeal
1068
            n_preds = [self.__get_num_preds(start_iteration, num_iteration, i, predict_type) for i in np.diff(sections)]
1069
            n_preds_sections = np.array([0] + n_preds, dtype=np.intp).cumsum()
1070
            preds = np.empty(sum(n_preds), dtype=np.float64)
1071
1072
            for (start_idx, end_idx), (start_idx_pred, end_idx_pred) in zip(zip(sections, sections[1:]),
                                                                            zip(n_preds_sections, n_preds_sections[1:])):
1073
                # avoid memory consumption by arrays concatenation operations
1074
                inner_predict(csr[start_idx:end_idx], start_iteration, num_iteration, predict_type, preds[start_idx_pred:end_idx_pred])
1075
1076
            return preds, nrow
        else:
1077
            return inner_predict(csr, start_iteration, num_iteration, predict_type)
Guolin Ke's avatar
Guolin Ke committed
1078

1079
    def __pred_for_csc(self, csc, start_iteration, num_iteration, predict_type):
1080
        """Predict for a CSC data."""
1081
        def inner_predict_sparse(csc, start_iteration, num_iteration, predict_type):
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
            ptr_indptr, type_ptr_indptr, __ = c_int_array(csc.indptr)
            ptr_data, type_ptr_data, _ = c_float_array(csc.data)
            csc_indices = csc.indices.astype(np.int32, copy=False)
            matrix_type = C_API_MATRIX_TYPE_CSC
            if type_ptr_indptr == C_API_DTYPE_INT32:
                out_ptr_indptr = ctypes.POINTER(ctypes.c_int32)()
            else:
                out_ptr_indptr = ctypes.POINTER(ctypes.c_int64)()
            out_ptr_indices = ctypes.POINTER(ctypes.c_int32)()
            if type_ptr_data == C_API_DTYPE_FLOAT32:
                out_ptr_data = ctypes.POINTER(ctypes.c_float)()
            else:
                out_ptr_data = ctypes.POINTER(ctypes.c_double)()
1095
            out_shape = np.empty(2, dtype=np.int64)
1096
1097
1098
            _safe_call(_LIB.LGBM_BoosterPredictSparseOutput(
                self.handle,
                ptr_indptr,
1099
                ctypes.c_int(type_ptr_indptr),
1100
1101
1102
1103
1104
1105
1106
                csc_indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
                ptr_data,
                ctypes.c_int(type_ptr_data),
                ctypes.c_int64(len(csc.indptr)),
                ctypes.c_int64(len(csc.data)),
                ctypes.c_int64(csc.shape[0]),
                ctypes.c_int(predict_type),
1107
                ctypes.c_int(start_iteration),
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
                ctypes.c_int(num_iteration),
                c_str(self.pred_parameter),
                ctypes.c_int(matrix_type),
                out_shape.ctypes.data_as(ctypes.POINTER(ctypes.c_int64)),
                ctypes.byref(out_ptr_indptr),
                ctypes.byref(out_ptr_indices),
                ctypes.byref(out_ptr_data)))
            matrices = self.__create_sparse_native(csc, out_shape, out_ptr_indptr, out_ptr_indices, out_ptr_data,
                                                   type_ptr_indptr, type_ptr_data, is_csr=False)
            nrow = csc.shape[0]
            return matrices, nrow

Guolin Ke's avatar
Guolin Ke committed
1120
        nrow = csc.shape[0]
1121
        if nrow > MAX_INT32:
1122
            return self.__pred_for_csr(csc.tocsr(), start_iteration, num_iteration, predict_type)
1123
        if predict_type == C_API_PREDICT_CONTRIB:
1124
1125
            return inner_predict_sparse(csc, start_iteration, num_iteration, predict_type)
        n_preds = self.__get_num_preds(start_iteration, num_iteration, nrow, predict_type)
1126
        preds = np.empty(n_preds, dtype=np.float64)
Guolin Ke's avatar
Guolin Ke committed
1127
1128
        out_num_preds = ctypes.c_int64(0)

1129
1130
        ptr_indptr, type_ptr_indptr, __ = c_int_array(csc.indptr)
        ptr_data, type_ptr_data, _ = c_float_array(csc.data)
Guolin Ke's avatar
Guolin Ke committed
1131

1132
        assert csc.shape[0] <= MAX_INT32
1133
        csc_indices = csc.indices.astype(np.int32, copy=False)
1134

Guolin Ke's avatar
Guolin Ke committed
1135
1136
1137
        _safe_call(_LIB.LGBM_BoosterPredictForCSC(
            self.handle,
            ptr_indptr,
1138
            ctypes.c_int(type_ptr_indptr),
1139
            csc_indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
Guolin Ke's avatar
Guolin Ke committed
1140
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
1141
1142
1143
1144
1145
            ctypes.c_int(type_ptr_data),
            ctypes.c_int64(len(csc.indptr)),
            ctypes.c_int64(len(csc.data)),
            ctypes.c_int64(csc.shape[0]),
            ctypes.c_int(predict_type),
1146
            ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
1147
            ctypes.c_int(num_iteration),
1148
            c_str(self.pred_parameter),
Guolin Ke's avatar
Guolin Ke committed
1149
            ctypes.byref(out_num_preds),
wxchan's avatar
wxchan committed
1150
            preds.ctypes.data_as(ctypes.POINTER(ctypes.c_double))))
wxchan's avatar
wxchan committed
1151
        if n_preds != out_num_preds.value:
1152
            raise ValueError("Wrong length for predict results")
wxchan's avatar
wxchan committed
1153
1154
        return preds, nrow

1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
    def current_iteration(self):
        """Get the index of the current iteration.

        Returns
        -------
        cur_iter : int
            The index of the current iteration.
        """
        out_cur_iter = ctypes.c_int(0)
        _safe_call(_LIB.LGBM_BoosterGetCurrentIteration(
            self.handle,
            ctypes.byref(out_cur_iter)))
        return out_cur_iter.value

wxchan's avatar
wxchan committed
1169

1170
class Dataset:
wxchan's avatar
wxchan committed
1171
    """Dataset in LightGBM."""
1172

1173
    def __init__(self, data, label=None, reference=None,
1174
                 weight=None, group=None, init_score=None, silent='warn',
1175
                 feature_name='auto', categorical_feature='auto', params=None,
wxchan's avatar
wxchan committed
1176
                 free_raw_data=True):
1177
        """Initialize Dataset.
1178

wxchan's avatar
wxchan committed
1179
1180
        Parameters
        ----------
1181
        data : str, pathlib.Path, numpy array, pandas DataFrame, H2O DataTable's Frame, scipy.sparse, Sequence, list of Sequence or list of numpy array
wxchan's avatar
wxchan committed
1182
            Data source of Dataset.
1183
            If str or pathlib.Path, it represents the path to a text file (CSV, TSV, or LibSVM) or a LightGBM Dataset binary file.
1184
        label : list, numpy 1-D array, pandas Series / one-column DataFrame or None, optional (default=None)
1185
1186
1187
            Label of the data.
        reference : Dataset or None, optional (default=None)
            If this is Dataset for validation, training data should be used as reference.
1188
        weight : list, numpy 1-D array, pandas Series or None, optional (default=None)
wxchan's avatar
wxchan committed
1189
            Weight for each instance.
1190
        group : list, numpy 1-D array, pandas Series or None, optional (default=None)
1191
1192
1193
            Group/query data.
            Only used in the learning-to-rank task.
            sum(group) = n_samples.
1194
1195
            For example, if you have a 100-document dataset with ``group = [10, 20, 40, 10, 10, 10]``, that means that you have 6 groups,
            where the first 10 records are in the first group, records 11-30 are in the second group, records 31-70 are in the third group, etc.
1196
        init_score : list, list of lists (for multi-class task), numpy array, pandas Series, pandas DataFrame (for multi-class task), or None, optional (default=None)
1197
            Init score for Dataset.
1198
1199
        silent : bool, optional (default=False)
            Whether to print messages during construction.
1200
        feature_name : list of str, or 'auto', optional (default="auto")
1201
1202
            Feature names.
            If 'auto' and data is pandas DataFrame, data columns names are used.
1203
        categorical_feature : list of str or int, or 'auto', optional (default="auto")
1204
1205
            Categorical features.
            If list of int, interpreted as indices.
1206
            If list of str, interpreted as feature names (need to specify ``feature_name`` as well).
1207
            If 'auto' and data is pandas DataFrame, pandas unordered categorical columns are used.
1208
            All values in categorical features should be less than int32 max value (2147483647).
1209
            Large values could be memory consuming. Consider using consecutive integers starting from zero.
1210
            All negative values in categorical features will be treated as missing values.
1211
            The output cannot be monotonically constrained with respect to a categorical feature.
Nikita Titov's avatar
Nikita Titov committed
1212
        params : dict or None, optional (default=None)
1213
            Other parameters for Dataset.
Nikita Titov's avatar
Nikita Titov committed
1214
        free_raw_data : bool, optional (default=True)
1215
            If True, raw data is freed after constructing inner Dataset.
wxchan's avatar
wxchan committed
1216
        """
wxchan's avatar
wxchan committed
1217
1218
1219
1220
1221
1222
        self.handle = None
        self.data = data
        self.label = label
        self.reference = reference
        self.weight = weight
        self.group = group
1223
        self.init_score = init_score
wxchan's avatar
wxchan committed
1224
1225
        self.silent = silent
        self.feature_name = feature_name
1226
        self.categorical_feature = categorical_feature
1227
        self.params = deepcopy(params)
wxchan's avatar
wxchan committed
1228
1229
        self.free_raw_data = free_raw_data
        self.used_indices = None
1230
        self.need_slice = True
wxchan's avatar
wxchan committed
1231
        self._predictor = None
1232
        self.pandas_categorical = None
1233
        self.params_back_up = None
1234
1235
        self.feature_penalty = None
        self.monotone_constraints = None
1236
        self.version = 0
1237
        self._start_row = 0  # Used when pushing rows one by one.
wxchan's avatar
wxchan committed
1238
1239

    def __del__(self):
1240
1241
1242
1243
        try:
            self._free_handle()
        except AttributeError:
            pass
1244

1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
    def _create_sample_indices(self, total_nrow: int) -> np.ndarray:
        """Get an array of randomly chosen indices from this ``Dataset``.

        Indices are sampled without replacement.

        Parameters
        ----------
        total_nrow : int
            Total number of rows to sample from.
            If this value is greater than the value of parameter ``bin_construct_sample_cnt``, only ``bin_construct_sample_cnt`` indices will be used.
            If Dataset has multiple input data, this should be the sum of rows of every file.

        Returns
        -------
        indices : numpy array
            Indices for sampled data.
        """
        param_str = param_dict_to_str(self.get_params())
        sample_cnt = _get_sample_count(total_nrow, param_str)
        indices = np.empty(sample_cnt, dtype=np.int32)
        ptr_data, _, _ = c_int_array(indices)
        actual_sample_cnt = ctypes.c_int32(0)

        _safe_call(_LIB.LGBM_SampleIndices(
            ctypes.c_int32(total_nrow),
            c_str(param_str),
            ptr_data,
            ctypes.byref(actual_sample_cnt),
        ))
1274
1275
        assert sample_cnt == actual_sample_cnt.value
        return indices
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310

    def _init_from_ref_dataset(self, total_nrow: int, ref_dataset: 'Dataset') -> 'Dataset':
        """Create dataset from a reference dataset.

        Parameters
        ----------
        total_nrow : int
            Number of rows expected to add to dataset.
        ref_dataset : Dataset
            Reference dataset to extract meta from.

        Returns
        -------
        self : Dataset
            Constructed Dataset object.
        """
        self.handle = ctypes.c_void_p()
        _safe_call(_LIB.LGBM_DatasetCreateByReference(
            ref_dataset,
            ctypes.c_int64(total_nrow),
            ctypes.byref(self.handle),
        ))
        return self

    def _init_from_sample(
        self,
        sample_data: List[np.ndarray],
        sample_indices: List[np.ndarray],
        sample_cnt: int,
        total_nrow: int,
    ) -> "Dataset":
        """Create Dataset from sampled data structures.

        Parameters
        ----------
1311
        sample_data : list of numpy array
1312
            Sample data for each column.
1313
        sample_indices : list of numpy array
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
            Sample data row index for each column.
        sample_cnt : int
            Number of samples.
        total_nrow : int
            Total number of rows for all input files.

        Returns
        -------
        self : Dataset
            Constructed Dataset object.
        """
        ncol = len(sample_indices)
        assert len(sample_data) == ncol, "#sample data column != #column indices"

        for i in range(ncol):
            if sample_data[i].dtype != np.double:
                raise ValueError(f"sample_data[{i}] type {sample_data[i].dtype} is not double")
            if sample_indices[i].dtype != np.int32:
                raise ValueError(f"sample_indices[{i}] type {sample_indices[i].dtype} is not int32")

        # c type: double**
        # each double* element points to start of each column of sample data.
        sample_col_ptr = (ctypes.POINTER(ctypes.c_double) * ncol)()
        # c type int**
        # each int* points to start of indices for each column
        indices_col_ptr = (ctypes.POINTER(ctypes.c_int32) * ncol)()
        for i in range(ncol):
            sample_col_ptr[i] = c_float_array(sample_data[i])[0]
            indices_col_ptr[i] = c_int_array(sample_indices[i])[0]

        num_per_col = np.array([len(d) for d in sample_indices], dtype=np.int32)
        num_per_col_ptr, _, _ = c_int_array(num_per_col)

        self.handle = ctypes.c_void_p()
        params_str = param_dict_to_str(self.get_params())
        _safe_call(_LIB.LGBM_DatasetCreateFromSampledColumn(
            ctypes.cast(sample_col_ptr, ctypes.POINTER(ctypes.POINTER(ctypes.c_double))),
            ctypes.cast(indices_col_ptr, ctypes.POINTER(ctypes.POINTER(ctypes.c_int32))),
            ctypes.c_int32(ncol),
            num_per_col_ptr,
            ctypes.c_int32(sample_cnt),
            ctypes.c_int32(total_nrow),
            c_str(params_str),
            ctypes.byref(self.handle),
        ))
        return self

    def _push_rows(self, data: np.ndarray) -> 'Dataset':
        """Add rows to Dataset.

        Parameters
        ----------
        data : numpy 1-D array
            New data to add to the Dataset.

        Returns
        -------
        self : Dataset
            Dataset object.
        """
        nrow, ncol = data.shape
        data = data.reshape(data.size)
        data_ptr, data_type, _ = c_float_array(data)

        _safe_call(_LIB.LGBM_DatasetPushRows(
            self.handle,
            data_ptr,
            data_type,
            ctypes.c_int32(nrow),
            ctypes.c_int32(ncol),
            ctypes.c_int32(self._start_row),
        ))
        self._start_row += nrow
        return self

1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
    def get_params(self):
        """Get the used parameters in the Dataset.

        Returns
        -------
        params : dict or None
            The used parameters in this Dataset object.
        """
        if self.params is not None:
            # no min_data, nthreads and verbose in this function
            dataset_params = _ConfigAliases.get("bin_construct_sample_cnt",
                                                "categorical_feature",
                                                "data_random_seed",
                                                "enable_bundle",
                                                "feature_pre_filter",
                                                "forcedbins_filename",
                                                "group_column",
                                                "header",
                                                "ignore_column",
                                                "is_enable_sparse",
                                                "label_column",
1410
                                                "linear_tree",
1411
1412
1413
1414
                                                "max_bin",
                                                "max_bin_by_feature",
                                                "min_data_in_bin",
                                                "pre_partition",
Nikita Titov's avatar
Nikita Titov committed
1415
                                                "precise_float_parser",
1416
1417
1418
1419
1420
1421
                                                "two_round",
                                                "use_missing",
                                                "weight_column",
                                                "zero_as_missing")
            return {k: v for k, v in self.params.items() if k in dataset_params}

1422
    def _free_handle(self):
1423
        if self.handle is not None:
1424
            _safe_call(_LIB.LGBM_DatasetFree(self.handle))
1425
            self.handle = None
Guolin Ke's avatar
Guolin Ke committed
1426
1427
1428
        self.need_slice = True
        if self.used_indices is not None:
            self.data = None
Nikita Titov's avatar
Nikita Titov committed
1429
        return self
wxchan's avatar
wxchan committed
1430

Guolin Ke's avatar
Guolin Ke committed
1431
1432
    def _set_init_score_by_predictor(self, predictor, data, used_indices=None):
        data_has_header = False
1433
        if isinstance(data, (str, Path)):
Guolin Ke's avatar
Guolin Ke committed
1434
            # check data has header or not
1435
            data_has_header = any(self.params.get(alias, False) for alias in _ConfigAliases.get("header"))
Guolin Ke's avatar
Guolin Ke committed
1436
        num_data = self.num_data()
1437
1438
1439
1440
1441
1442
1443
        if predictor is not None:
            init_score = predictor.predict(data,
                                           raw_score=True,
                                           data_has_header=data_has_header,
                                           is_reshape=False)
            if used_indices is not None:
                assert not self.need_slice
1444
                if isinstance(data, (str, Path)):
1445
                    sub_init_score = np.empty(num_data * predictor.num_class, dtype=np.float64)
1446
                    assert num_data == len(used_indices)
1447
1448
                    for i in range(len(used_indices)):
                        for j in range(predictor.num_class):
1449
1450
1451
1452
                            sub_init_score[i * predictor.num_class + j] = init_score[used_indices[i] * predictor.num_class + j]
                    init_score = sub_init_score
            if predictor.num_class > 1:
                # need to regroup init_score
1453
                new_init_score = np.empty(init_score.size, dtype=np.float64)
1454
1455
                for i in range(num_data):
                    for j in range(predictor.num_class):
1456
1457
1458
                        new_init_score[j * num_data + i] = init_score[i * predictor.num_class + j]
                init_score = new_init_score
        elif self.init_score is not None:
1459
            init_score = np.zeros(self.init_score.shape, dtype=np.float64)
1460
1461
        else:
            return self
Guolin Ke's avatar
Guolin Ke committed
1462
1463
        self.set_init_score(init_score)

1464
    def _lazy_init(self, data, label=None, reference=None,
1465
                   weight=None, group=None, init_score=None, predictor=None,
wxchan's avatar
wxchan committed
1466
                   silent=False, feature_name='auto',
1467
                   categorical_feature='auto', params=None):
wxchan's avatar
wxchan committed
1468
1469
        if data is None:
            self.handle = None
Nikita Titov's avatar
Nikita Titov committed
1470
            return self
Guolin Ke's avatar
Guolin Ke committed
1471
1472
1473
        if reference is not None:
            self.pandas_categorical = reference.pandas_categorical
            categorical_feature = reference.categorical_feature
1474
1475
1476
1477
        data, feature_name, categorical_feature, self.pandas_categorical = _data_from_pandas(data,
                                                                                             feature_name,
                                                                                             categorical_feature,
                                                                                             self.pandas_categorical)
wxchan's avatar
wxchan committed
1478
        label = _label_from_pandas(label)
Guolin Ke's avatar
Guolin Ke committed
1479

1480
        # process for args
wxchan's avatar
wxchan committed
1481
        params = {} if params is None else params
1482
1483
1484
        args_names = (getattr(self.__class__, '_lazy_init')
                      .__code__
                      .co_varnames[:getattr(self.__class__, '_lazy_init').__code__.co_argcount])
1485
        for key in params.keys():
1486
            if key in args_names:
1487
1488
                _log_warning(f'{key} keyword has been found in `params` and will be ignored.\n'
                             f'Please use {key} argument of the Dataset constructor to pass this parameter.')
1489
        # user can set verbose with params, it has higher priority
1490
1491
1492
1493
1494
        if silent != "warn":
            _log_warning("'silent' argument is deprecated and will be removed in a future release of LightGBM. "
                         "Pass 'verbose' parameter via 'params' instead.")
        else:
            silent = False
1495
        if not any(verbose_alias in params for verbose_alias in _ConfigAliases.get("verbosity")) and silent:
1496
            params["verbose"] = -1
1497
        # get categorical features
1498
1499
1500
1501
1502
1503
        if categorical_feature is not None:
            categorical_indices = set()
            feature_dict = {}
            if feature_name is not None:
                feature_dict = {name: i for i, name in enumerate(feature_name)}
            for name in categorical_feature:
1504
                if isinstance(name, str) and name in feature_dict:
1505
                    categorical_indices.add(feature_dict[name])
1506
                elif isinstance(name, int):
1507
1508
                    categorical_indices.add(name)
                else:
1509
                    raise TypeError(f"Wrong type({type(name).__name__}) or unknown name({name}) in categorical_feature")
1510
            if categorical_indices:
1511
1512
                for cat_alias in _ConfigAliases.get("categorical_feature"):
                    if cat_alias in params:
1513
1514
1515
                        # If the params[cat_alias] is equal to categorical_indices, do not report the warning.
                        if not(isinstance(params[cat_alias], list) and set(params[cat_alias]) == categorical_indices):
                            _log_warning(f'{cat_alias} in param dict is overridden.')
1516
                        params.pop(cat_alias, None)
1517
                params['categorical_column'] = sorted(categorical_indices)
1518

wxchan's avatar
wxchan committed
1519
        params_str = param_dict_to_str(params)
1520
        self.params = params
1521
        # process for reference dataset
wxchan's avatar
wxchan committed
1522
        ref_dataset = None
wxchan's avatar
wxchan committed
1523
        if isinstance(reference, Dataset):
1524
            ref_dataset = reference.construct().handle
wxchan's avatar
wxchan committed
1525
1526
        elif reference is not None:
            raise TypeError('Reference dataset should be None or dataset instance')
1527
        # start construct data
1528
        if isinstance(data, (str, Path)):
wxchan's avatar
wxchan committed
1529
1530
            self.handle = ctypes.c_void_p()
            _safe_call(_LIB.LGBM_DatasetCreateFromFile(
1531
                c_str(str(data)),
wxchan's avatar
wxchan committed
1532
1533
1534
1535
1536
                c_str(params_str),
                ref_dataset,
                ctypes.byref(self.handle)))
        elif isinstance(data, scipy.sparse.csr_matrix):
            self.__init_from_csr(data, params_str, ref_dataset)
Guolin Ke's avatar
Guolin Ke committed
1537
1538
        elif isinstance(data, scipy.sparse.csc_matrix):
            self.__init_from_csc(data, params_str, ref_dataset)
wxchan's avatar
wxchan committed
1539
1540
        elif isinstance(data, np.ndarray):
            self.__init_from_np2d(data, params_str, ref_dataset)
1541
1542
1543
1544
1545
1546
1547
1548
1549
        elif isinstance(data, list) and len(data) > 0:
            if all(isinstance(x, np.ndarray) for x in data):
                self.__init_from_list_np2d(data, params_str, ref_dataset)
            elif all(isinstance(x, Sequence) for x in data):
                self.__init_from_seqs(data, ref_dataset)
            else:
                raise TypeError('Data list can only be of ndarray or Sequence')
        elif isinstance(data, Sequence):
            self.__init_from_seqs([data], ref_dataset)
1550
        elif isinstance(data, dt_DataTable):
1551
            self.__init_from_np2d(data.to_numpy(), params_str, ref_dataset)
wxchan's avatar
wxchan committed
1552
1553
1554
1555
        else:
            try:
                csr = scipy.sparse.csr_matrix(data)
                self.__init_from_csr(csr, params_str, ref_dataset)
1556
            except BaseException:
1557
                raise TypeError(f'Cannot initialize Dataset from {type(data).__name__}')
wxchan's avatar
wxchan committed
1558
1559
1560
        if label is not None:
            self.set_label(label)
        if self.get_label() is None:
1561
            raise ValueError("Label should not be None")
wxchan's avatar
wxchan committed
1562
1563
1564
1565
        if weight is not None:
            self.set_weight(weight)
        if group is not None:
            self.set_group(group)
1566
1567
        if isinstance(predictor, _InnerPredictor):
            if self._predictor is None and init_score is not None:
1568
                _log_warning("The init_score will be overridden by the prediction of init_model.")
Guolin Ke's avatar
Guolin Ke committed
1569
            self._set_init_score_by_predictor(predictor, data)
1570
1571
        elif init_score is not None:
            self.set_init_score(init_score)
Guolin Ke's avatar
Guolin Ke committed
1572
        elif predictor is not None:
1573
            raise TypeError(f'Wrong predictor type {type(predictor).__name__}')
Guolin Ke's avatar
Guolin Ke committed
1574
        # set feature names
Nikita Titov's avatar
Nikita Titov committed
1575
        return self.set_feature_name(feature_name)
wxchan's avatar
wxchan committed
1576

1577
1578
    @staticmethod
    def _yield_row_from_seqlist(seqs: List[Sequence], indices: Iterable[int]):
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
        offset = 0
        seq_id = 0
        seq = seqs[seq_id]
        for row_id in indices:
            assert row_id >= offset, "sample indices are expected to be monotonic"
            while row_id >= offset + len(seq):
                offset += len(seq)
                seq_id += 1
                seq = seqs[seq_id]
            id_in_seq = row_id - offset
            row = seq[id_in_seq]
            yield row if row.flags['OWNDATA'] else row.copy()

    def __sample(self, seqs: List[Sequence], total_nrow: int) -> Tuple[List[np.ndarray], List[np.ndarray]]:
        """Sample data from seqs.

        Mimics behavior in c_api.cpp:LGBM_DatasetCreateFromMats()

        Returns
        -------
            sampled_rows, sampled_row_indices
        """
        indices = self._create_sample_indices(total_nrow)

        # Select sampled rows, transpose to column order.
1604
        sampled = np.array([row for row in self._yield_row_from_seqlist(seqs, indices)])
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
        sampled = sampled.T

        filtered = []
        filtered_idx = []
        sampled_row_range = np.arange(len(indices), dtype=np.int32)
        for col in sampled:
            col_predicate = (np.abs(col) > ZERO_THRESHOLD) | np.isnan(col)
            filtered_col = col[col_predicate]
            filtered_row_idx = sampled_row_range[col_predicate]

            filtered.append(filtered_col)
            filtered_idx.append(filtered_row_idx)

        return filtered, filtered_idx

    def __init_from_seqs(self, seqs: List[Sequence], ref_dataset: Optional['Dataset'] = None):
        """
        Initialize data from list of Sequence objects.

        Sequence: Generic Data Access Object
            Supports random access and access by batch if properly defined by user

        Data scheme uniformity are trusted, not checked
        """
        total_nrow = sum(len(seq) for seq in seqs)

        # create validation dataset from ref_dataset
        if ref_dataset is not None:
            self._init_from_ref_dataset(total_nrow, ref_dataset)
        else:
            param_str = param_dict_to_str(self.get_params())
            sample_cnt = _get_sample_count(total_nrow, param_str)

            sample_data, col_indices = self.__sample(seqs, total_nrow)
            self._init_from_sample(sample_data, col_indices, sample_cnt, total_nrow)

        for seq in seqs:
            nrow = len(seq)
            batch_size = getattr(seq, 'batch_size', None) or Sequence.batch_size
            for start in range(0, nrow, batch_size):
                end = min(start + batch_size, nrow)
                self._push_rows(seq[start:end])
        return self

wxchan's avatar
wxchan committed
1649
    def __init_from_np2d(self, mat, params_str, ref_dataset):
1650
        """Initialize data from a 2-D numpy matrix."""
wxchan's avatar
wxchan committed
1651
1652
1653
1654
1655
1656
        if len(mat.shape) != 2:
            raise ValueError('Input numpy.ndarray must be 2 dimensional')

        self.handle = ctypes.c_void_p()
        if mat.dtype == np.float32 or mat.dtype == np.float64:
            data = np.array(mat.reshape(mat.size), dtype=mat.dtype, copy=False)
1657
        else:  # change non-float data to float data, need to copy
wxchan's avatar
wxchan committed
1658
1659
            data = np.array(mat.reshape(mat.size), dtype=np.float32)

1660
        ptr_data, type_ptr_data, _ = c_float_array(data)
wxchan's avatar
wxchan committed
1661
1662
        _safe_call(_LIB.LGBM_DatasetCreateFromMat(
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
1663
            ctypes.c_int(type_ptr_data),
1664
1665
            ctypes.c_int32(mat.shape[0]),
            ctypes.c_int32(mat.shape[1]),
Guolin Ke's avatar
Guolin Ke committed
1666
            ctypes.c_int(C_API_IS_ROW_MAJOR),
wxchan's avatar
wxchan committed
1667
1668
1669
            c_str(params_str),
            ref_dataset,
            ctypes.byref(self.handle)))
Nikita Titov's avatar
Nikita Titov committed
1670
        return self
wxchan's avatar
wxchan committed
1671

1672
    def __init_from_list_np2d(self, mats, params_str, ref_dataset):
1673
        """Initialize data from a list of 2-D numpy matrices."""
1674
        ncol = mats[0].shape[1]
1675
        nrow = np.empty((len(mats),), np.int32)
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
        if mats[0].dtype == np.float64:
            ptr_data = (ctypes.POINTER(ctypes.c_double) * len(mats))()
        else:
            ptr_data = (ctypes.POINTER(ctypes.c_float) * len(mats))()

        holders = []
        type_ptr_data = None

        for i, mat in enumerate(mats):
            if len(mat.shape) != 2:
                raise ValueError('Input numpy.ndarray must be 2 dimensional')

            if mat.shape[1] != ncol:
                raise ValueError('Input arrays must have same number of columns')

            nrow[i] = mat.shape[0]

            if mat.dtype == np.float32 or mat.dtype == np.float64:
                mats[i] = np.array(mat.reshape(mat.size), dtype=mat.dtype, copy=False)
1695
            else:  # change non-float data to float data, need to copy
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
                mats[i] = np.array(mat.reshape(mat.size), dtype=np.float32)

            chunk_ptr_data, chunk_type_ptr_data, holder = c_float_array(mats[i])
            if type_ptr_data is not None and chunk_type_ptr_data != type_ptr_data:
                raise ValueError('Input chunks must have same type')
            ptr_data[i] = chunk_ptr_data
            type_ptr_data = chunk_type_ptr_data
            holders.append(holder)

        self.handle = ctypes.c_void_p()
        _safe_call(_LIB.LGBM_DatasetCreateFromMats(
1707
            ctypes.c_int32(len(mats)),
1708
1709
1710
            ctypes.cast(ptr_data, ctypes.POINTER(ctypes.POINTER(ctypes.c_double))),
            ctypes.c_int(type_ptr_data),
            nrow.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
1711
            ctypes.c_int32(ncol),
1712
1713
1714
1715
            ctypes.c_int(C_API_IS_ROW_MAJOR),
            c_str(params_str),
            ref_dataset,
            ctypes.byref(self.handle)))
Nikita Titov's avatar
Nikita Titov committed
1716
        return self
1717

wxchan's avatar
wxchan committed
1718
    def __init_from_csr(self, csr, params_str, ref_dataset):
1719
        """Initialize data from a CSR matrix."""
wxchan's avatar
wxchan committed
1720
        if len(csr.indices) != len(csr.data):
1721
            raise ValueError(f'Length mismatch: {len(csr.indices)} vs {len(csr.data)}')
wxchan's avatar
wxchan committed
1722
1723
        self.handle = ctypes.c_void_p()

1724
1725
        ptr_indptr, type_ptr_indptr, __ = c_int_array(csr.indptr)
        ptr_data, type_ptr_data, _ = c_float_array(csr.data)
wxchan's avatar
wxchan committed
1726

1727
        assert csr.shape[1] <= MAX_INT32
1728
        csr_indices = csr.indices.astype(np.int32, copy=False)
1729

wxchan's avatar
wxchan committed
1730
1731
        _safe_call(_LIB.LGBM_DatasetCreateFromCSR(
            ptr_indptr,
Guolin Ke's avatar
Guolin Ke committed
1732
            ctypes.c_int(type_ptr_indptr),
1733
            csr_indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
wxchan's avatar
wxchan committed
1734
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
1735
1736
1737
1738
            ctypes.c_int(type_ptr_data),
            ctypes.c_int64(len(csr.indptr)),
            ctypes.c_int64(len(csr.data)),
            ctypes.c_int64(csr.shape[1]),
wxchan's avatar
wxchan committed
1739
1740
1741
            c_str(params_str),
            ref_dataset,
            ctypes.byref(self.handle)))
Nikita Titov's avatar
Nikita Titov committed
1742
        return self
wxchan's avatar
wxchan committed
1743

Guolin Ke's avatar
Guolin Ke committed
1744
    def __init_from_csc(self, csc, params_str, ref_dataset):
1745
        """Initialize data from a CSC matrix."""
Guolin Ke's avatar
Guolin Ke committed
1746
        if len(csc.indices) != len(csc.data):
1747
            raise ValueError(f'Length mismatch: {len(csc.indices)} vs {len(csc.data)}')
Guolin Ke's avatar
Guolin Ke committed
1748
1749
        self.handle = ctypes.c_void_p()

1750
1751
        ptr_indptr, type_ptr_indptr, __ = c_int_array(csc.indptr)
        ptr_data, type_ptr_data, _ = c_float_array(csc.data)
Guolin Ke's avatar
Guolin Ke committed
1752

1753
        assert csc.shape[0] <= MAX_INT32
1754
        csc_indices = csc.indices.astype(np.int32, copy=False)
1755

Guolin Ke's avatar
Guolin Ke committed
1756
1757
        _safe_call(_LIB.LGBM_DatasetCreateFromCSC(
            ptr_indptr,
Guolin Ke's avatar
Guolin Ke committed
1758
            ctypes.c_int(type_ptr_indptr),
1759
            csc_indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
Guolin Ke's avatar
Guolin Ke committed
1760
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
1761
1762
1763
1764
            ctypes.c_int(type_ptr_data),
            ctypes.c_int64(len(csc.indptr)),
            ctypes.c_int64(len(csc.data)),
            ctypes.c_int64(csc.shape[0]),
Guolin Ke's avatar
Guolin Ke committed
1765
1766
1767
            c_str(params_str),
            ref_dataset,
            ctypes.byref(self.handle)))
Nikita Titov's avatar
Nikita Titov committed
1768
        return self
Guolin Ke's avatar
Guolin Ke committed
1769

1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
    @staticmethod
    def _compare_params_for_warning(params, other_params):
        """Compare params.

        It is only for the warning purpose. Thus some keys are ignored.

        Returns
        -------
        compare_result: bool
          If they are equal, return True; Otherwise, return False.
        """
        ignore_keys = _ConfigAliases.get("categorical_feature")
        if params is None:
            params = {}
        if other_params is None:
            other_params = {}
        for k in other_params:
            if k not in ignore_keys:
                if k not in params or params[k] != other_params[k]:
                    return False
        for k in params:
            if k not in ignore_keys:
                if k not in other_params or params[k] != other_params[k]:
                    return False
        return True

wxchan's avatar
wxchan committed
1796
    def construct(self):
1797
1798
1799
1800
1801
        """Lazy init.

        Returns
        -------
        self : Dataset
Nikita Titov's avatar
Nikita Titov committed
1802
            Constructed Dataset object.
1803
        """
1804
        if self.handle is None:
wxchan's avatar
wxchan committed
1805
            if self.reference is not None:
1806
                reference_params = self.reference.get_params()
1807
1808
1809
1810
                params = self.get_params()
                if params != reference_params:
                    if self._compare_params_for_warning(params, reference_params) is False:
                        _log_warning('Overriding the parameters from Reference Dataset.')
1811
                    self._update_params(reference_params)
wxchan's avatar
wxchan committed
1812
                if self.used_indices is None:
1813
                    # create valid
1814
                    self._lazy_init(self.data, label=self.label, reference=self.reference,
1815
1816
                                    weight=self.weight, group=self.group,
                                    init_score=self.init_score, predictor=self._predictor,
1817
                                    silent=self.silent, feature_name=self.feature_name, params=self.params)
wxchan's avatar
wxchan committed
1818
                else:
1819
                    # construct subset
wxchan's avatar
wxchan committed
1820
                    used_indices = list_to_1d_numpy(self.used_indices, np.int32, name='used_indices')
1821
                    assert used_indices.flags.c_contiguous
Guolin Ke's avatar
Guolin Ke committed
1822
                    if self.reference.group is not None:
1823
                        group_info = np.array(self.reference.group).astype(np.int32, copy=False)
1824
                        _, self.group = np.unique(np.repeat(range(len(group_info)), repeats=group_info)[self.used_indices],
1825
                                                  return_counts=True)
1826
                    self.handle = ctypes.c_void_p()
wxchan's avatar
wxchan committed
1827
1828
                    params_str = param_dict_to_str(self.params)
                    _safe_call(_LIB.LGBM_DatasetGetSubset(
1829
                        self.reference.construct().handle,
wxchan's avatar
wxchan committed
1830
                        used_indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
1831
                        ctypes.c_int32(used_indices.shape[0]),
wxchan's avatar
wxchan committed
1832
1833
                        c_str(params_str),
                        ctypes.byref(self.handle)))
Guolin Ke's avatar
Guolin Ke committed
1834
1835
                    if not self.free_raw_data:
                        self.get_data()
Guolin Ke's avatar
Guolin Ke committed
1836
1837
                    if self.group is not None:
                        self.set_group(self.group)
wxchan's avatar
wxchan committed
1838
1839
                    if self.get_label() is None:
                        raise ValueError("Label should not be None.")
Guolin Ke's avatar
Guolin Ke committed
1840
1841
1842
                    if isinstance(self._predictor, _InnerPredictor) and self._predictor is not self.reference._predictor:
                        self.get_data()
                        self._set_init_score_by_predictor(self._predictor, self.data, used_indices)
wxchan's avatar
wxchan committed
1843
            else:
1844
                # create train
1845
                self._lazy_init(self.data, label=self.label,
1846
1847
1848
                                weight=self.weight, group=self.group,
                                init_score=self.init_score, predictor=self._predictor,
                                silent=self.silent, feature_name=self.feature_name,
1849
                                categorical_feature=self.categorical_feature, params=self.params)
wxchan's avatar
wxchan committed
1850
1851
1852
            if self.free_raw_data:
                self.data = None
        return self
wxchan's avatar
wxchan committed
1853

wxchan's avatar
wxchan committed
1854
    def create_valid(self, data, label=None, weight=None, group=None,
1855
                     init_score=None, silent='warn', params=None):
1856
        """Create validation data align with current Dataset.
wxchan's avatar
wxchan committed
1857
1858
1859

        Parameters
        ----------
1860
        data : str, pathlib.Path, numpy array, pandas DataFrame, H2O DataTable's Frame, scipy.sparse, Sequence, list of Sequence or list of numpy array
wxchan's avatar
wxchan committed
1861
            Data source of Dataset.
1862
            If str or pathlib.Path, it represents the path to a text file (CSV, TSV, or LibSVM) or a LightGBM Dataset binary file.
1863
        label : list, numpy 1-D array, pandas Series / one-column DataFrame or None, optional (default=None)
1864
1865
            Label of the data.
        weight : list, numpy 1-D array, pandas Series or None, optional (default=None)
wxchan's avatar
wxchan committed
1866
            Weight for each instance.
1867
        group : list, numpy 1-D array, pandas Series or None, optional (default=None)
1868
1869
1870
            Group/query data.
            Only used in the learning-to-rank task.
            sum(group) = n_samples.
1871
1872
            For example, if you have a 100-document dataset with ``group = [10, 20, 40, 10, 10, 10]``, that means that you have 6 groups,
            where the first 10 records are in the first group, records 11-30 are in the second group, records 31-70 are in the third group, etc.
1873
        init_score : list, list of lists (for multi-class task), numpy array, pandas Series, pandas DataFrame (for multi-class task), or None, optional (default=None)
1874
            Init score for Dataset.
1875
1876
        silent : bool, optional (default=False)
            Whether to print messages during construction.
Nikita Titov's avatar
Nikita Titov committed
1877
        params : dict or None, optional (default=None)
1878
            Other parameters for validation Dataset.
1879
1880
1881

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1882
1883
        valid : Dataset
            Validation Dataset with reference to self.
wxchan's avatar
wxchan committed
1884
        """
1885
        ret = Dataset(data, label=label, reference=self,
1886
1887
                      weight=weight, group=group, init_score=init_score,
                      silent=silent, params=params, free_raw_data=self.free_raw_data)
wxchan's avatar
wxchan committed
1888
        ret._predictor = self._predictor
1889
        ret.pandas_categorical = self.pandas_categorical
wxchan's avatar
wxchan committed
1890
        return ret
wxchan's avatar
wxchan committed
1891

wxchan's avatar
wxchan committed
1892
    def subset(self, used_indices, params=None):
1893
        """Get subset of current Dataset.
wxchan's avatar
wxchan committed
1894
1895
1896
1897

        Parameters
        ----------
        used_indices : list of int
1898
            Indices used to create the subset.
Nikita Titov's avatar
Nikita Titov committed
1899
        params : dict or None, optional (default=None)
1900
            These parameters will be passed to Dataset constructor.
1901
1902
1903
1904
1905

        Returns
        -------
        subset : Dataset
            Subset of the current Dataset.
wxchan's avatar
wxchan committed
1906
        """
wxchan's avatar
wxchan committed
1907
1908
        if params is None:
            params = self.params
wxchan's avatar
wxchan committed
1909
        ret = Dataset(None, reference=self, feature_name=self.feature_name,
1910
1911
                      categorical_feature=self.categorical_feature, params=params,
                      free_raw_data=self.free_raw_data)
wxchan's avatar
wxchan committed
1912
        ret._predictor = self._predictor
1913
        ret.pandas_categorical = self.pandas_categorical
1914
        ret.used_indices = sorted(used_indices)
wxchan's avatar
wxchan committed
1915
1916
1917
        return ret

    def save_binary(self, filename):
1918
        """Save Dataset to a binary file.
wxchan's avatar
wxchan committed
1919

1920
1921
1922
1923
1924
        .. note::

            Please note that `init_score` is not saved in binary file.
            If you need it, please set it again after loading Dataset.

wxchan's avatar
wxchan committed
1925
1926
        Parameters
        ----------
1927
        filename : str or pathlib.Path
wxchan's avatar
wxchan committed
1928
            Name of the output file.
Nikita Titov's avatar
Nikita Titov committed
1929
1930
1931
1932
1933

        Returns
        -------
        self : Dataset
            Returns self.
wxchan's avatar
wxchan committed
1934
1935
1936
        """
        _safe_call(_LIB.LGBM_DatasetSaveBinary(
            self.construct().handle,
1937
            c_str(str(filename))))
Nikita Titov's avatar
Nikita Titov committed
1938
        return self
wxchan's avatar
wxchan committed
1939
1940

    def _update_params(self, params):
1941
1942
        if not params:
            return self
1943
        params = deepcopy(params)
1944
1945
1946
1947
1948

        def update():
            if not self.params:
                self.params = params
            else:
1949
                self.params_back_up = deepcopy(self.params)
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
                self.params.update(params)

        if self.handle is None:
            update()
        elif params is not None:
            ret = _LIB.LGBM_DatasetUpdateParamChecking(
                c_str(param_dict_to_str(self.params)),
                c_str(param_dict_to_str(params)))
            if ret != 0:
                # could be updated if data is not freed
                if self.data is not None:
                    update()
                    self._free_handle()
                else:
1964
                    raise LightGBMError(_LIB.LGBM_GetLastError().decode('utf-8'))
Nikita Titov's avatar
Nikita Titov committed
1965
        return self
wxchan's avatar
wxchan committed
1966

1967
    def _reverse_update_params(self):
1968
        if self.handle is None:
1969
            self.params = deepcopy(self.params_back_up)
1970
            self.params_back_up = None
Nikita Titov's avatar
Nikita Titov committed
1971
        return self
1972

wxchan's avatar
wxchan committed
1973
    def set_field(self, field_name, data):
wxchan's avatar
wxchan committed
1974
        """Set property into the Dataset.
wxchan's avatar
wxchan committed
1975
1976
1977

        Parameters
        ----------
1978
        field_name : str
1979
            The field name of the information.
1980
1981
        data : list, list of lists (for multi-class task), numpy array, pandas Series, pandas DataFrame (for multi-class task), or None
            The data to be set.
Nikita Titov's avatar
Nikita Titov committed
1982
1983
1984
1985
1986

        Returns
        -------
        self : Dataset
            Dataset with set property.
wxchan's avatar
wxchan committed
1987
        """
1988
        if self.handle is None:
1989
            raise Exception(f"Cannot set {field_name} before construct dataset")
wxchan's avatar
wxchan committed
1990
        if data is None:
1991
            # set to None
wxchan's avatar
wxchan committed
1992
1993
1994
1995
            _safe_call(_LIB.LGBM_DatasetSetField(
                self.handle,
                c_str(field_name),
                None,
Guolin Ke's avatar
Guolin Ke committed
1996
1997
                ctypes.c_int(0),
                ctypes.c_int(FIELD_TYPE_MAPPER[field_name])))
Nikita Titov's avatar
Nikita Titov committed
1998
            return self
1999
        if field_name == 'init_score':
Guolin Ke's avatar
Guolin Ke committed
2000
            dtype = np.float64
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
            if _is_1d_collection(data):
                data = list_to_1d_numpy(data, dtype, name=field_name)
            elif _is_2d_collection(data):
                data = _data_to_2d_numpy(data, dtype, name=field_name)
                data = data.ravel(order='F')
            else:
                raise TypeError(
                    'init_score must be list, numpy 1-D array or pandas Series.\n'
                    'In multiclass classification init_score can also be a list of lists, numpy 2-D array or pandas DataFrame.'
                )
        else:
            dtype = np.int32 if field_name == 'group' else np.float32
            data = list_to_1d_numpy(data, dtype, name=field_name)

2015
2016
        if data.dtype == np.float32 or data.dtype == np.float64:
            ptr_data, type_data, _ = c_float_array(data)
wxchan's avatar
wxchan committed
2017
        elif data.dtype == np.int32:
2018
            ptr_data, type_data, _ = c_int_array(data)
wxchan's avatar
wxchan committed
2019
        else:
2020
            raise TypeError(f"Expected np.float32/64 or np.int32, met type({data.dtype})")
wxchan's avatar
wxchan committed
2021
        if type_data != FIELD_TYPE_MAPPER[field_name]:
2022
            raise TypeError("Input type error for set_field")
wxchan's avatar
wxchan committed
2023
2024
2025
2026
        _safe_call(_LIB.LGBM_DatasetSetField(
            self.handle,
            c_str(field_name),
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
2027
2028
            ctypes.c_int(len(data)),
            ctypes.c_int(type_data)))
2029
        self.version += 1
Nikita Titov's avatar
Nikita Titov committed
2030
        return self
wxchan's avatar
wxchan committed
2031

wxchan's avatar
wxchan committed
2032
2033
    def get_field(self, field_name):
        """Get property from the Dataset.
wxchan's avatar
wxchan committed
2034
2035
2036

        Parameters
        ----------
2037
        field_name : str
2038
            The field name of the information.
wxchan's avatar
wxchan committed
2039
2040
2041

        Returns
        -------
2042
        info : numpy array or None
2043
            A numpy array with information from the Dataset.
Guolin Ke's avatar
Guolin Ke committed
2044
        """
2045
        if self.handle is None:
2046
            raise Exception(f"Cannot get {field_name} before construct Dataset")
2047
2048
        tmp_out_len = ctypes.c_int(0)
        out_type = ctypes.c_int(0)
wxchan's avatar
wxchan committed
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
        ret = ctypes.POINTER(ctypes.c_void_p)()
        _safe_call(_LIB.LGBM_DatasetGetField(
            self.handle,
            c_str(field_name),
            ctypes.byref(tmp_out_len),
            ctypes.byref(ret),
            ctypes.byref(out_type)))
        if out_type.value != FIELD_TYPE_MAPPER[field_name]:
            raise TypeError("Return type error for get_field")
        if tmp_out_len.value == 0:
            return None
        if out_type.value == C_API_DTYPE_INT32:
2061
            arr = cint32_array_to_numpy(ctypes.cast(ret, ctypes.POINTER(ctypes.c_int32)), tmp_out_len.value)
wxchan's avatar
wxchan committed
2062
        elif out_type.value == C_API_DTYPE_FLOAT32:
2063
            arr = cfloat32_array_to_numpy(ctypes.cast(ret, ctypes.POINTER(ctypes.c_float)), tmp_out_len.value)
Guolin Ke's avatar
Guolin Ke committed
2064
        elif out_type.value == C_API_DTYPE_FLOAT64:
2065
            arr = cfloat64_array_to_numpy(ctypes.cast(ret, ctypes.POINTER(ctypes.c_double)), tmp_out_len.value)
2066
        else:
wxchan's avatar
wxchan committed
2067
            raise TypeError("Unknown type")
2068
2069
2070
2071
2072
2073
        if field_name == 'init_score':
            num_data = self.num_data()
            num_classes = arr.size // num_data
            if num_classes > 1:
                arr = arr.reshape((num_data, num_classes), order='F')
        return arr
Guolin Ke's avatar
Guolin Ke committed
2074

2075
    def set_categorical_feature(self, categorical_feature):
2076
        """Set categorical features.
2077
2078
2079

        Parameters
        ----------
2080
        categorical_feature : list of int or str
2081
            Names or indices of categorical features.
Nikita Titov's avatar
Nikita Titov committed
2082
2083
2084
2085
2086

        Returns
        -------
        self : Dataset
            Dataset with set categorical features.
2087
2088
        """
        if self.categorical_feature == categorical_feature:
Nikita Titov's avatar
Nikita Titov committed
2089
            return self
2090
        if self.data is not None:
2091
2092
            if self.categorical_feature is None:
                self.categorical_feature = categorical_feature
Nikita Titov's avatar
Nikita Titov committed
2093
                return self._free_handle()
2094
            elif categorical_feature == 'auto':
Nikita Titov's avatar
Nikita Titov committed
2095
                return self
2096
            else:
2097
2098
2099
                if self.categorical_feature != 'auto':
                    _log_warning('categorical_feature in Dataset is overridden.\n'
                                 f'New categorical_feature is {sorted(list(categorical_feature))}')
2100
                self.categorical_feature = categorical_feature
Nikita Titov's avatar
Nikita Titov committed
2101
                return self._free_handle()
2102
        else:
2103
2104
            raise LightGBMError("Cannot set categorical feature after freed raw data, "
                                "set free_raw_data=False when construct Dataset to avoid this.")
2105

Guolin Ke's avatar
Guolin Ke committed
2106
    def _set_predictor(self, predictor):
2107
2108
2109
2110
        """Set predictor for continued training.

        It is not recommended for user to call this function.
        Please use init_model argument in engine.train() or engine.cv() instead.
Guolin Ke's avatar
Guolin Ke committed
2111
        """
2112
        if predictor is self._predictor and (predictor is None or predictor.current_iteration() == self._predictor.current_iteration()):
Nikita Titov's avatar
Nikita Titov committed
2113
            return self
2114
        if self.handle is None:
Guolin Ke's avatar
Guolin Ke committed
2115
            self._predictor = predictor
2116
2117
2118
2119
2120
2121
        elif self.data is not None:
            self._predictor = predictor
            self._set_init_score_by_predictor(self._predictor, self.data)
        elif self.used_indices is not None and self.reference is not None and self.reference.data is not None:
            self._predictor = predictor
            self._set_init_score_by_predictor(self._predictor, self.reference.data, self.used_indices)
Guolin Ke's avatar
Guolin Ke committed
2122
        else:
2123
2124
            raise LightGBMError("Cannot set predictor after freed raw data, "
                                "set free_raw_data=False when construct Dataset to avoid this.")
2125
        return self
Guolin Ke's avatar
Guolin Ke committed
2126
2127

    def set_reference(self, reference):
2128
        """Set reference Dataset.
Guolin Ke's avatar
Guolin Ke committed
2129
2130
2131
2132

        Parameters
        ----------
        reference : Dataset
2133
            Reference that is used as a template to construct the current Dataset.
Nikita Titov's avatar
Nikita Titov committed
2134
2135
2136
2137
2138

        Returns
        -------
        self : Dataset
            Dataset with set reference.
Guolin Ke's avatar
Guolin Ke committed
2139
        """
2140
2141
2142
        self.set_categorical_feature(reference.categorical_feature) \
            .set_feature_name(reference.feature_name) \
            ._set_predictor(reference._predictor)
2143
        # we're done if self and reference share a common upstream reference
2144
        if self.get_ref_chain().intersection(reference.get_ref_chain()):
Nikita Titov's avatar
Nikita Titov committed
2145
            return self
Guolin Ke's avatar
Guolin Ke committed
2146
2147
        if self.data is not None:
            self.reference = reference
Nikita Titov's avatar
Nikita Titov committed
2148
            return self._free_handle()
Guolin Ke's avatar
Guolin Ke committed
2149
        else:
2150
2151
            raise LightGBMError("Cannot set reference after freed raw data, "
                                "set free_raw_data=False when construct Dataset to avoid this.")
Guolin Ke's avatar
Guolin Ke committed
2152
2153

    def set_feature_name(self, feature_name):
2154
        """Set feature name.
Guolin Ke's avatar
Guolin Ke committed
2155
2156
2157

        Parameters
        ----------
2158
        feature_name : list of str
2159
            Feature names.
Nikita Titov's avatar
Nikita Titov committed
2160
2161
2162
2163
2164

        Returns
        -------
        self : Dataset
            Dataset with set feature name.
Guolin Ke's avatar
Guolin Ke committed
2165
        """
2166
2167
        if feature_name != 'auto':
            self.feature_name = feature_name
2168
        if self.handle is not None and feature_name is not None and feature_name != 'auto':
wxchan's avatar
wxchan committed
2169
            if len(feature_name) != self.num_feature():
2170
                raise ValueError(f"Length of feature_name({len(feature_name)}) and num_feature({self.num_feature()}) don't match")
2171
            c_feature_name = [c_str(name) for name in feature_name]
wxchan's avatar
wxchan committed
2172
2173
2174
            _safe_call(_LIB.LGBM_DatasetSetFeatureNames(
                self.handle,
                c_array(ctypes.c_char_p, c_feature_name),
Guolin Ke's avatar
Guolin Ke committed
2175
                ctypes.c_int(len(feature_name))))
Nikita Titov's avatar
Nikita Titov committed
2176
        return self
Guolin Ke's avatar
Guolin Ke committed
2177
2178

    def set_label(self, label):
2179
        """Set label of Dataset.
Guolin Ke's avatar
Guolin Ke committed
2180
2181
2182

        Parameters
        ----------
2183
        label : list, numpy 1-D array, pandas Series / one-column DataFrame or None
2184
            The label information to be set into Dataset.
Nikita Titov's avatar
Nikita Titov committed
2185
2186
2187
2188
2189

        Returns
        -------
        self : Dataset
            Dataset with set label.
Guolin Ke's avatar
Guolin Ke committed
2190
2191
        """
        self.label = label
2192
        if self.handle is not None:
2193
            label = list_to_1d_numpy(_label_from_pandas(label), name='label')
wxchan's avatar
wxchan committed
2194
            self.set_field('label', label)
2195
            self.label = self.get_field('label')  # original values can be modified at cpp side
Nikita Titov's avatar
Nikita Titov committed
2196
        return self
Guolin Ke's avatar
Guolin Ke committed
2197
2198

    def set_weight(self, weight):
2199
        """Set weight of each instance.
Guolin Ke's avatar
Guolin Ke committed
2200
2201
2202

        Parameters
        ----------
2203
        weight : list, numpy 1-D array, pandas Series or None
2204
            Weight to be set for each data point.
Nikita Titov's avatar
Nikita Titov committed
2205
2206
2207
2208
2209

        Returns
        -------
        self : Dataset
            Dataset with set weight.
Guolin Ke's avatar
Guolin Ke committed
2210
        """
2211
2212
        if weight is not None and np.all(weight == 1):
            weight = None
Guolin Ke's avatar
Guolin Ke committed
2213
        self.weight = weight
2214
        if self.handle is not None and weight is not None:
wxchan's avatar
wxchan committed
2215
2216
            weight = list_to_1d_numpy(weight, name='weight')
            self.set_field('weight', weight)
2217
            self.weight = self.get_field('weight')  # original values can be modified at cpp side
Nikita Titov's avatar
Nikita Titov committed
2218
        return self
Guolin Ke's avatar
Guolin Ke committed
2219
2220

    def set_init_score(self, init_score):
2221
        """Set init score of Booster to start from.
Guolin Ke's avatar
Guolin Ke committed
2222
2223
2224

        Parameters
        ----------
2225
        init_score : list, list of lists (for multi-class task), numpy array, pandas Series, pandas DataFrame (for multi-class task), or None
2226
            Init score for Booster.
Nikita Titov's avatar
Nikita Titov committed
2227
2228
2229
2230
2231

        Returns
        -------
        self : Dataset
            Dataset with set init score.
Guolin Ke's avatar
Guolin Ke committed
2232
2233
        """
        self.init_score = init_score
2234
        if self.handle is not None and init_score is not None:
wxchan's avatar
wxchan committed
2235
            self.set_field('init_score', init_score)
2236
            self.init_score = self.get_field('init_score')  # original values can be modified at cpp side
Nikita Titov's avatar
Nikita Titov committed
2237
        return self
Guolin Ke's avatar
Guolin Ke committed
2238
2239

    def set_group(self, group):
2240
        """Set group size of Dataset (used for ranking).
Guolin Ke's avatar
Guolin Ke committed
2241
2242
2243

        Parameters
        ----------
2244
        group : list, numpy 1-D array, pandas Series or None
2245
2246
2247
            Group/query data.
            Only used in the learning-to-rank task.
            sum(group) = n_samples.
2248
2249
            For example, if you have a 100-document dataset with ``group = [10, 20, 40, 10, 10, 10]``, that means that you have 6 groups,
            where the first 10 records are in the first group, records 11-30 are in the second group, records 31-70 are in the third group, etc.
Nikita Titov's avatar
Nikita Titov committed
2250
2251
2252
2253
2254

        Returns
        -------
        self : Dataset
            Dataset with set group.
Guolin Ke's avatar
Guolin Ke committed
2255
2256
        """
        self.group = group
2257
        if self.handle is not None and group is not None:
wxchan's avatar
wxchan committed
2258
2259
            group = list_to_1d_numpy(group, np.int32, name='group')
            self.set_field('group', group)
Nikita Titov's avatar
Nikita Titov committed
2260
        return self
Guolin Ke's avatar
Guolin Ke committed
2261

2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
    def get_feature_name(self):
        """Get the names of columns (features) in the Dataset.

        Returns
        -------
        feature_names : list
            The names of columns (features) in the Dataset.
        """
        if self.handle is None:
            raise LightGBMError("Cannot get feature_name before construct dataset")
        num_feature = self.num_feature()
        tmp_out_len = ctypes.c_int(0)
        reserved_string_buffer_size = 255
        required_string_buffer_size = ctypes.c_size_t(0)
2276
        string_buffers = [ctypes.create_string_buffer(reserved_string_buffer_size) for _ in range(num_feature)]
2277
2278
2279
        ptr_string_buffers = (ctypes.c_char_p * num_feature)(*map(ctypes.addressof, string_buffers))
        _safe_call(_LIB.LGBM_DatasetGetFeatureNames(
            self.handle,
2280
            ctypes.c_int(num_feature),
2281
            ctypes.byref(tmp_out_len),
2282
            ctypes.c_size_t(reserved_string_buffer_size),
2283
2284
2285
2286
            ctypes.byref(required_string_buffer_size),
            ptr_string_buffers))
        if num_feature != tmp_out_len.value:
            raise ValueError("Length of feature names doesn't equal with num_feature")
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
        actual_string_buffer_size = required_string_buffer_size.value
        # if buffer length is not long enough, reallocate buffers
        if reserved_string_buffer_size < actual_string_buffer_size:
            string_buffers = [ctypes.create_string_buffer(actual_string_buffer_size) for _ in range(num_feature)]
            ptr_string_buffers = (ctypes.c_char_p * num_feature)(*map(ctypes.addressof, string_buffers))
            _safe_call(_LIB.LGBM_DatasetGetFeatureNames(
                self.handle,
                ctypes.c_int(num_feature),
                ctypes.byref(tmp_out_len),
                ctypes.c_size_t(actual_string_buffer_size),
                ctypes.byref(required_string_buffer_size),
                ptr_string_buffers))
2299
        return [string_buffers[i].value.decode('utf-8') for i in range(num_feature)]
2300

Guolin Ke's avatar
Guolin Ke committed
2301
    def get_label(self):
2302
        """Get the label of the Dataset.
Guolin Ke's avatar
Guolin Ke committed
2303
2304
2305

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
2306
        label : numpy array or None
2307
            The label information from the Dataset.
Guolin Ke's avatar
Guolin Ke committed
2308
        """
2309
        if self.label is None:
wxchan's avatar
wxchan committed
2310
            self.label = self.get_field('label')
Guolin Ke's avatar
Guolin Ke committed
2311
2312
2313
        return self.label

    def get_weight(self):
2314
        """Get the weight of the Dataset.
Guolin Ke's avatar
Guolin Ke committed
2315
2316
2317

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
2318
        weight : numpy array or None
2319
            Weight for each data point from the Dataset.
Guolin Ke's avatar
Guolin Ke committed
2320
        """
2321
        if self.weight is None:
wxchan's avatar
wxchan committed
2322
            self.weight = self.get_field('weight')
Guolin Ke's avatar
Guolin Ke committed
2323
2324
2325
        return self.weight

    def get_init_score(self):
2326
        """Get the initial score of the Dataset.
Guolin Ke's avatar
Guolin Ke committed
2327
2328
2329

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
2330
        init_score : numpy array or None
2331
            Init score of Booster.
Guolin Ke's avatar
Guolin Ke committed
2332
        """
2333
        if self.init_score is None:
wxchan's avatar
wxchan committed
2334
            self.init_score = self.get_field('init_score')
Guolin Ke's avatar
Guolin Ke committed
2335
2336
        return self.init_score

2337
2338
2339
2340
2341
    def get_data(self):
        """Get the raw data of the Dataset.

        Returns
        -------
2342
        data : str, pathlib.Path, numpy array, pandas DataFrame, H2O DataTable's Frame, scipy.sparse, Sequence, list of Sequence or list of numpy array or None
2343
2344
2345
2346
            Raw data used in the Dataset construction.
        """
        if self.handle is None:
            raise Exception("Cannot get data before construct Dataset")
Guolin Ke's avatar
Guolin Ke committed
2347
2348
2349
2350
2351
        if self.need_slice and self.used_indices is not None and self.reference is not None:
            self.data = self.reference.data
            if self.data is not None:
                if isinstance(self.data, np.ndarray) or scipy.sparse.issparse(self.data):
                    self.data = self.data[self.used_indices, :]
2352
                elif isinstance(self.data, pd_DataFrame):
Guolin Ke's avatar
Guolin Ke committed
2353
                    self.data = self.data.iloc[self.used_indices].copy()
2354
                elif isinstance(self.data, dt_DataTable):
Guolin Ke's avatar
Guolin Ke committed
2355
                    self.data = self.data[self.used_indices, :]
2356
2357
2358
2359
                elif isinstance(self.data, Sequence):
                    self.data = self.data[self.used_indices]
                elif isinstance(self.data, list) and len(self.data) > 0 and all(isinstance(x, Sequence) for x in self.data):
                    self.data = np.array([row for row in self._yield_row_from_seqlist(self.data, self.used_indices)])
Guolin Ke's avatar
Guolin Ke committed
2360
                else:
2361
2362
                    _log_warning(f"Cannot subset {type(self.data).__name__} type of raw data.\n"
                                 "Returning original raw data")
2363
            self.need_slice = False
Guolin Ke's avatar
Guolin Ke committed
2364
2365
2366
        if self.data is None:
            raise LightGBMError("Cannot call `get_data` after freed raw data, "
                                "set free_raw_data=False when construct Dataset to avoid this.")
2367
2368
        return self.data

Guolin Ke's avatar
Guolin Ke committed
2369
    def get_group(self):
2370
        """Get the group of the Dataset.
Guolin Ke's avatar
Guolin Ke committed
2371
2372
2373

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
2374
        group : numpy array or None
2375
2376
2377
            Group/query data.
            Only used in the learning-to-rank task.
            sum(group) = n_samples.
2378
2379
            For example, if you have a 100-document dataset with ``group = [10, 20, 40, 10, 10, 10]``, that means that you have 6 groups,
            where the first 10 records are in the first group, records 11-30 are in the second group, records 31-70 are in the third group, etc.
Guolin Ke's avatar
Guolin Ke committed
2380
        """
2381
        if self.group is None:
wxchan's avatar
wxchan committed
2382
            self.group = self.get_field('group')
Guolin Ke's avatar
Guolin Ke committed
2383
2384
            if self.group is not None:
                # group data from LightGBM is boundaries data, need to convert to group size
Nikita Titov's avatar
Nikita Titov committed
2385
                self.group = np.diff(self.group)
Guolin Ke's avatar
Guolin Ke committed
2386
2387
2388
        return self.group

    def num_data(self):
2389
        """Get the number of rows in the Dataset.
Guolin Ke's avatar
Guolin Ke committed
2390
2391
2392

        Returns
        -------
2393
2394
        number_of_rows : int
            The number of rows in the Dataset.
Guolin Ke's avatar
Guolin Ke committed
2395
        """
2396
        if self.handle is not None:
2397
            ret = ctypes.c_int(0)
wxchan's avatar
wxchan committed
2398
2399
2400
            _safe_call(_LIB.LGBM_DatasetGetNumData(self.handle,
                                                   ctypes.byref(ret)))
            return ret.value
Guolin Ke's avatar
Guolin Ke committed
2401
        else:
2402
            raise LightGBMError("Cannot get num_data before construct dataset")
Guolin Ke's avatar
Guolin Ke committed
2403
2404

    def num_feature(self):
2405
        """Get the number of columns (features) in the Dataset.
Guolin Ke's avatar
Guolin Ke committed
2406
2407
2408

        Returns
        -------
2409
2410
        number_of_columns : int
            The number of columns (features) in the Dataset.
Guolin Ke's avatar
Guolin Ke committed
2411
        """
2412
        if self.handle is not None:
2413
            ret = ctypes.c_int(0)
wxchan's avatar
wxchan committed
2414
2415
2416
            _safe_call(_LIB.LGBM_DatasetGetNumFeature(self.handle,
                                                      ctypes.byref(ret)))
            return ret.value
Guolin Ke's avatar
Guolin Ke committed
2417
        else:
2418
            raise LightGBMError("Cannot get num_feature before construct dataset")
Guolin Ke's avatar
Guolin Ke committed
2419

2420
    def get_ref_chain(self, ref_limit=100):
2421
2422
2423
2424
2425
        """Get a chain of Dataset objects.

        Starts with r, then goes to r.reference (if exists),
        then to r.reference.reference, etc.
        until we hit ``ref_limit`` or a reference loop.
2426
2427
2428
2429
2430

        Parameters
        ----------
        ref_limit : int, optional (default=100)
            The limit number of references.
2431
2432
2433

        Returns
        -------
2434
2435
2436
        ref_chain : set of Dataset
            Chain of references of the Datasets.
        """
2437
        head = self
2438
        ref_chain = set()
2439
2440
        while len(ref_chain) < ref_limit:
            if isinstance(head, Dataset):
2441
                ref_chain.add(head)
2442
2443
2444
2445
2446
2447
                if (head.reference is not None) and (head.reference not in ref_chain):
                    head = head.reference
                else:
                    break
            else:
                break
Nikita Titov's avatar
Nikita Titov committed
2448
        return ref_chain
2449

2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
    def add_features_from(self, other):
        """Add features from other Dataset to the current Dataset.

        Both Datasets must be constructed before calling this method.

        Parameters
        ----------
        other : Dataset
            The Dataset to take features from.

        Returns
        -------
        self : Dataset
            Dataset with the new features added.
        """
        if self.handle is None or other.handle is None:
            raise ValueError('Both source and target Datasets must be constructed before adding features')
        _safe_call(_LIB.LGBM_DatasetAddFeaturesFrom(self.handle, other.handle))
Guolin Ke's avatar
Guolin Ke committed
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
        was_none = self.data is None
        old_self_data_type = type(self.data).__name__
        if other.data is None:
            self.data = None
        elif self.data is not None:
            if isinstance(self.data, np.ndarray):
                if isinstance(other.data, np.ndarray):
                    self.data = np.hstack((self.data, other.data))
                elif scipy.sparse.issparse(other.data):
                    self.data = np.hstack((self.data, other.data.toarray()))
2478
                elif isinstance(other.data, pd_DataFrame):
Guolin Ke's avatar
Guolin Ke committed
2479
                    self.data = np.hstack((self.data, other.data.values))
2480
                elif isinstance(other.data, dt_DataTable):
Guolin Ke's avatar
Guolin Ke committed
2481
2482
2483
2484
2485
2486
2487
                    self.data = np.hstack((self.data, other.data.to_numpy()))
                else:
                    self.data = None
            elif scipy.sparse.issparse(self.data):
                sparse_format = self.data.getformat()
                if isinstance(other.data, np.ndarray) or scipy.sparse.issparse(other.data):
                    self.data = scipy.sparse.hstack((self.data, other.data), format=sparse_format)
2488
                elif isinstance(other.data, pd_DataFrame):
Guolin Ke's avatar
Guolin Ke committed
2489
                    self.data = scipy.sparse.hstack((self.data, other.data.values), format=sparse_format)
2490
                elif isinstance(other.data, dt_DataTable):
Guolin Ke's avatar
Guolin Ke committed
2491
2492
2493
                    self.data = scipy.sparse.hstack((self.data, other.data.to_numpy()), format=sparse_format)
                else:
                    self.data = None
2494
            elif isinstance(self.data, pd_DataFrame):
Guolin Ke's avatar
Guolin Ke committed
2495
2496
                if not PANDAS_INSTALLED:
                    raise LightGBMError("Cannot add features to DataFrame type of raw data "
2497
2498
                                        "without pandas installed. "
                                        "Install pandas and restart your session.")
Guolin Ke's avatar
Guolin Ke committed
2499
                if isinstance(other.data, np.ndarray):
2500
                    self.data = concat((self.data, pd_DataFrame(other.data)),
Guolin Ke's avatar
Guolin Ke committed
2501
2502
                                       axis=1, ignore_index=True)
                elif scipy.sparse.issparse(other.data):
2503
                    self.data = concat((self.data, pd_DataFrame(other.data.toarray())),
Guolin Ke's avatar
Guolin Ke committed
2504
                                       axis=1, ignore_index=True)
2505
                elif isinstance(other.data, pd_DataFrame):
Guolin Ke's avatar
Guolin Ke committed
2506
2507
                    self.data = concat((self.data, other.data),
                                       axis=1, ignore_index=True)
2508
2509
                elif isinstance(other.data, dt_DataTable):
                    self.data = concat((self.data, pd_DataFrame(other.data.to_numpy())),
Guolin Ke's avatar
Guolin Ke committed
2510
2511
2512
                                       axis=1, ignore_index=True)
                else:
                    self.data = None
2513
            elif isinstance(self.data, dt_DataTable):
Guolin Ke's avatar
Guolin Ke committed
2514
                if isinstance(other.data, np.ndarray):
2515
                    self.data = dt_DataTable(np.hstack((self.data.to_numpy(), other.data)))
Guolin Ke's avatar
Guolin Ke committed
2516
                elif scipy.sparse.issparse(other.data):
2517
2518
2519
2520
2521
                    self.data = dt_DataTable(np.hstack((self.data.to_numpy(), other.data.toarray())))
                elif isinstance(other.data, pd_DataFrame):
                    self.data = dt_DataTable(np.hstack((self.data.to_numpy(), other.data.values)))
                elif isinstance(other.data, dt_DataTable):
                    self.data = dt_DataTable(np.hstack((self.data.to_numpy(), other.data.to_numpy())))
Guolin Ke's avatar
Guolin Ke committed
2522
2523
2524
2525
2526
                else:
                    self.data = None
            else:
                self.data = None
        if self.data is None:
2527
2528
            err_msg = (f"Cannot add features from {type(other.data).__name__} type of raw data to "
                       f"{old_self_data_type} type of raw data.\n")
Guolin Ke's avatar
Guolin Ke committed
2529
2530
            err_msg += ("Set free_raw_data=False when construct Dataset to avoid this"
                        if was_none else "Freeing raw data")
2531
            _log_warning(err_msg)
Guolin Ke's avatar
Guolin Ke committed
2532
        self.feature_name = self.get_feature_name()
2533
2534
        _log_warning("Reseting categorical features.\n"
                     "You can set new categorical features via ``set_categorical_feature`` method")
Guolin Ke's avatar
Guolin Ke committed
2535
2536
        self.categorical_feature = "auto"
        self.pandas_categorical = None
2537
2538
        return self

2539
    def _dump_text(self, filename):
2540
2541
2542
2543
2544
2545
        """Save Dataset to a text file.

        This format cannot be loaded back in by LightGBM, but is useful for debugging purposes.

        Parameters
        ----------
2546
        filename : str or pathlib.Path
2547
2548
2549
2550
2551
2552
2553
2554
2555
            Name of the output file.

        Returns
        -------
        self : Dataset
            Returns self.
        """
        _safe_call(_LIB.LGBM_DatasetDumpText(
            self.construct().handle,
2556
            c_str(str(filename))))
2557
2558
        return self

wxchan's avatar
wxchan committed
2559

2560
class Booster:
2561
    """Booster in LightGBM."""
2562

2563
    def __init__(self, params=None, train_set=None, model_file=None, model_str=None, silent='warn'):
2564
        """Initialize the Booster.
wxchan's avatar
wxchan committed
2565
2566
2567

        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
2568
        params : dict or None, optional (default=None)
2569
2570
2571
            Parameters for Booster.
        train_set : Dataset or None, optional (default=None)
            Training dataset.
2572
        model_file : str, pathlib.Path or None, optional (default=None)
wxchan's avatar
wxchan committed
2573
            Path to the model file.
2574
        model_str : str or None, optional (default=None)
2575
            Model will be loaded from this string.
2576
2577
        silent : bool, optional (default=False)
            Whether to print messages during construction.
wxchan's avatar
wxchan committed
2578
        """
2579
        self.handle = None
2580
        self.network = False
wxchan's avatar
wxchan committed
2581
        self.__need_reload_eval_info = True
2582
        self._train_data_name = "training"
wxchan's avatar
wxchan committed
2583
        self.__attr = {}
2584
        self.__set_objective_to_none = False
wxchan's avatar
wxchan committed
2585
        self.best_iteration = -1
wxchan's avatar
wxchan committed
2586
        self.best_score = {}
2587
        params = {} if params is None else deepcopy(params)
2588
        # user can set verbose with params, it has higher priority
2589
2590
2591
2592
2593
        if silent != 'warn':
            _log_warning("'silent' argument is deprecated and will be removed in a future release of LightGBM. "
                         "Pass 'verbose' parameter via 'params' instead.")
        else:
            silent = False
2594
        if not any(verbose_alias in params for verbose_alias in _ConfigAliases.get("verbosity")) and silent:
2595
            params["verbose"] = -1
wxchan's avatar
wxchan committed
2596
        if train_set is not None:
2597
            # Training task
wxchan's avatar
wxchan committed
2598
            if not isinstance(train_set, Dataset):
2599
                raise TypeError(f'Training data should be Dataset instance, met {type(train_set).__name__}')
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
            params = _choose_param_value(
                main_param_name="machines",
                params=params,
                default_value=None
            )
            # if "machines" is given, assume user wants to do distributed learning, and set up network
            if params["machines"] is None:
                params.pop("machines", None)
            else:
                machines = params["machines"]
                if isinstance(machines, str):
                    num_machines_from_machine_list = len(machines.split(','))
                elif isinstance(machines, (list, set)):
                    num_machines_from_machine_list = len(machines)
                    machines = ','.join(machines)
                else:
                    raise ValueError("Invalid machines in params.")

                params = _choose_param_value(
                    main_param_name="num_machines",
                    params=params,
                    default_value=num_machines_from_machine_list
                )
                params = _choose_param_value(
                    main_param_name="local_listen_port",
                    params=params,
                    default_value=12400
                )
                self.set_network(
                    machines=machines,
                    local_listen_port=params["local_listen_port"],
                    listen_time_out=params.get("time_out", 120),
                    num_machines=params["num_machines"]
                )
2634
            # construct booster object
2635
2636
2637
2638
            train_set.construct()
            # copy the parameters from train_set
            params.update(train_set.get_params())
            params_str = param_dict_to_str(params)
2639
            self.handle = ctypes.c_void_p()
wxchan's avatar
wxchan committed
2640
            _safe_call(_LIB.LGBM_BoosterCreate(
2641
                train_set.handle,
wxchan's avatar
wxchan committed
2642
2643
                c_str(params_str),
                ctypes.byref(self.handle)))
2644
            # save reference to data
wxchan's avatar
wxchan committed
2645
2646
2647
2648
            self.train_set = train_set
            self.valid_sets = []
            self.name_valid_sets = []
            self.__num_dataset = 1
Guolin Ke's avatar
Guolin Ke committed
2649
2650
            self.__init_predictor = train_set._predictor
            if self.__init_predictor is not None:
wxchan's avatar
wxchan committed
2651
2652
                _safe_call(_LIB.LGBM_BoosterMerge(
                    self.handle,
Guolin Ke's avatar
Guolin Ke committed
2653
                    self.__init_predictor.handle))
Guolin Ke's avatar
Guolin Ke committed
2654
            out_num_class = ctypes.c_int(0)
wxchan's avatar
wxchan committed
2655
2656
2657
2658
            _safe_call(_LIB.LGBM_BoosterGetNumClasses(
                self.handle,
                ctypes.byref(out_num_class)))
            self.__num_class = out_num_class.value
2659
            # buffer for inner predict
wxchan's avatar
wxchan committed
2660
2661
2662
            self.__inner_predict_buffer = [None]
            self.__is_predicted_cur_iter = [False]
            self.__get_eval_info()
2663
            self.pandas_categorical = train_set.pandas_categorical
2664
            self.train_set_version = train_set.version
wxchan's avatar
wxchan committed
2665
        elif model_file is not None:
2666
            # Prediction task
Guolin Ke's avatar
Guolin Ke committed
2667
            out_num_iterations = ctypes.c_int(0)
2668
            self.handle = ctypes.c_void_p()
wxchan's avatar
wxchan committed
2669
            _safe_call(_LIB.LGBM_BoosterCreateFromModelfile(
2670
                c_str(str(model_file)),
wxchan's avatar
wxchan committed
2671
2672
                ctypes.byref(out_num_iterations),
                ctypes.byref(self.handle)))
Guolin Ke's avatar
Guolin Ke committed
2673
            out_num_class = ctypes.c_int(0)
wxchan's avatar
wxchan committed
2674
2675
2676
2677
            _safe_call(_LIB.LGBM_BoosterGetNumClasses(
                self.handle,
                ctypes.byref(out_num_class)))
            self.__num_class = out_num_class.value
2678
            self.pandas_categorical = _load_pandas_categorical(file_name=model_file)
2679
        elif model_str is not None:
2680
            self.model_from_string(model_str, verbose="_silent_false")
wxchan's avatar
wxchan committed
2681
        else:
2682
2683
            raise TypeError('Need at least one training dataset or model file or model string '
                            'to create Booster instance')
2684
        self.params = params
wxchan's avatar
wxchan committed
2685
2686

    def __del__(self):
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
        try:
            if self.network:
                self.free_network()
        except AttributeError:
            pass
        try:
            if self.handle is not None:
                _safe_call(_LIB.LGBM_BoosterFree(self.handle))
        except AttributeError:
            pass
wxchan's avatar
wxchan committed
2697

wxchan's avatar
wxchan committed
2698
2699
2700
2701
    def __copy__(self):
        return self.__deepcopy__(None)

    def __deepcopy__(self, _):
2702
        model_str = self.model_to_string(num_iteration=-1)
2703
        booster = Booster(model_str=model_str)
2704
        return booster
wxchan's avatar
wxchan committed
2705
2706
2707
2708
2709
2710
2711

    def __getstate__(self):
        this = self.__dict__.copy()
        handle = this['handle']
        this.pop('train_set', None)
        this.pop('valid_sets', None)
        if handle is not None:
2712
            this["handle"] = self.model_to_string(num_iteration=-1)
wxchan's avatar
wxchan committed
2713
2714
2715
        return this

    def __setstate__(self, state):
2716
2717
        model_str = state.get('handle', None)
        if model_str is not None:
wxchan's avatar
wxchan committed
2718
            handle = ctypes.c_void_p()
Guolin Ke's avatar
Guolin Ke committed
2719
            out_num_iterations = ctypes.c_int(0)
2720
2721
2722
2723
            _safe_call(_LIB.LGBM_BoosterLoadModelFromString(
                c_str(model_str),
                ctypes.byref(out_num_iterations),
                ctypes.byref(handle)))
wxchan's avatar
wxchan committed
2724
2725
2726
            state['handle'] = handle
        self.__dict__.update(state)

wxchan's avatar
wxchan committed
2727
    def free_dataset(self):
Nikita Titov's avatar
Nikita Titov committed
2728
2729
2730
2731
2732
2733
2734
        """Free Booster's Datasets.

        Returns
        -------
        self : Booster
            Booster without Datasets.
        """
wxchan's avatar
wxchan committed
2735
2736
        self.__dict__.pop('train_set', None)
        self.__dict__.pop('valid_sets', None)
2737
        self.__num_dataset = 0
Nikita Titov's avatar
Nikita Titov committed
2738
        return self
wxchan's avatar
wxchan committed
2739

2740
2741
2742
    def _free_buffer(self):
        self.__inner_predict_buffer = []
        self.__is_predicted_cur_iter = []
Nikita Titov's avatar
Nikita Titov committed
2743
        return self
2744

2745
2746
2747
2748
2749
2750
2751
    def set_network(
        self,
        machines: Union[List[str], Set[str], str],
        local_listen_port: int = 12400,
        listen_time_out: int = 120,
        num_machines: int = 1
    ) -> "Booster":
2752
2753
2754
2755
        """Set the network configuration.

        Parameters
        ----------
2756
        machines : list, set or str
2757
            Names of machines.
Nikita Titov's avatar
Nikita Titov committed
2758
        local_listen_port : int, optional (default=12400)
2759
            TCP listen port for local machines.
Nikita Titov's avatar
Nikita Titov committed
2760
        listen_time_out : int, optional (default=120)
2761
            Socket time-out in minutes.
Nikita Titov's avatar
Nikita Titov committed
2762
        num_machines : int, optional (default=1)
2763
            The number of machines for distributed learning application.
Nikita Titov's avatar
Nikita Titov committed
2764
2765
2766
2767
2768

        Returns
        -------
        self : Booster
            Booster with set network.
2769
        """
2770
2771
        if isinstance(machines, (list, set)):
            machines = ','.join(machines)
2772
2773
2774
2775
2776
        _safe_call(_LIB.LGBM_NetworkInit(c_str(machines),
                                         ctypes.c_int(local_listen_port),
                                         ctypes.c_int(listen_time_out),
                                         ctypes.c_int(num_machines)))
        self.network = True
Nikita Titov's avatar
Nikita Titov committed
2777
        return self
2778
2779

    def free_network(self):
Nikita Titov's avatar
Nikita Titov committed
2780
2781
2782
2783
2784
2785
2786
        """Free Booster's network.

        Returns
        -------
        self : Booster
            Booster with freed network.
        """
2787
2788
        _safe_call(_LIB.LGBM_NetworkFree())
        self.network = False
Nikita Titov's avatar
Nikita Titov committed
2789
        return self
2790

2791
2792
2793
    def trees_to_dataframe(self):
        """Parse the fitted model and return in an easy-to-read pandas DataFrame.

2794
2795
2796
2797
        The returned DataFrame has the following columns.

            - ``tree_index`` : int64, which tree a node belongs to. 0-based, so a value of ``6``, for example, means "this node is in the 7th tree".
            - ``node_depth`` : int64, how far a node is from the root of the tree. The root node has a value of ``1``, its direct children are ``2``, etc.
2798
2799
2800
2801
2802
            - ``node_index`` : str, unique identifier for a node.
            - ``left_child`` : str, ``node_index`` of the child node to the left of a split. ``None`` for leaf nodes.
            - ``right_child`` : str, ``node_index`` of the child node to the right of a split. ``None`` for leaf nodes.
            - ``parent_index`` : str, ``node_index`` of this node's parent. ``None`` for the root node.
            - ``split_feature`` : str, name of the feature used for splitting. ``None`` for leaf nodes.
2803
2804
            - ``split_gain`` : float64, gain from adding this split to the tree. ``NaN`` for leaf nodes.
            - ``threshold`` : float64, value of the feature used to decide which side of the split a record will go down. ``NaN`` for leaf nodes.
2805
            - ``decision_type`` : str, logical operator describing how to compare a value to ``threshold``.
2806
2807
              For example, ``split_feature = "Column_10", threshold = 15, decision_type = "<="`` means that
              records where ``Column_10 <= 15`` follow the left side of the split, otherwise follows the right side of the split. ``None`` for leaf nodes.
2808
2809
            - ``missing_direction`` : str, split direction that missing values should go to. ``None`` for leaf nodes.
            - ``missing_type`` : str, describes what types of values are treated as missing.
2810
2811
2812
2813
            - ``value`` : float64, predicted value for this leaf node, multiplied by the learning rate.
            - ``weight`` : float64 or int64, sum of hessian (second-order derivative of objective), summed over observations that fall in this node.
            - ``count`` : int64, number of records in the training data that fall into this node.

2814
2815
2816
2817
2818
2819
        Returns
        -------
        result : pandas DataFrame
            Returns a pandas DataFrame of the parsed model.
        """
        if not PANDAS_INSTALLED:
2820
2821
            raise LightGBMError('This method cannot be run without pandas installed. '
                                'You must install pandas and restart your session to use this method.')
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832

        if self.num_trees() == 0:
            raise LightGBMError('There are no trees in this Booster and thus nothing to parse')

        def _is_split_node(tree):
            return 'split_index' in tree.keys()

        def create_node_record(tree, node_depth=1, tree_index=None,
                               feature_names=None, parent_node=None):

            def _get_node_index(tree, tree_index):
2833
                tree_num = f'{tree_index}-' if tree_index is not None else ''
2834
2835
2836
                is_split = _is_split_node(tree)
                node_type = 'S' if is_split else 'L'
                # if a single node tree it won't have `leaf_index` so return 0
2837
2838
                node_num = tree.get('split_index' if is_split else 'leaf_index', 0)
                return f"{tree_num}{node_type}{node_num}"
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850

            def _get_split_feature(tree, feature_names):
                if _is_split_node(tree):
                    if feature_names is not None:
                        feature_name = feature_names[tree['split_feature']]
                    else:
                        feature_name = tree['split_feature']
                else:
                    feature_name = None
                return feature_name

            def _is_single_node_tree(tree):
2851
                return set(tree.keys()) == {'leaf_value'}
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924

            # Create the node record, and populate universal data members
            node = OrderedDict()
            node['tree_index'] = tree_index
            node['node_depth'] = node_depth
            node['node_index'] = _get_node_index(tree, tree_index)
            node['left_child'] = None
            node['right_child'] = None
            node['parent_index'] = parent_node
            node['split_feature'] = _get_split_feature(tree, feature_names)
            node['split_gain'] = None
            node['threshold'] = None
            node['decision_type'] = None
            node['missing_direction'] = None
            node['missing_type'] = None
            node['value'] = None
            node['weight'] = None
            node['count'] = None

            # Update values to reflect node type (leaf or split)
            if _is_split_node(tree):
                node['left_child'] = _get_node_index(tree['left_child'], tree_index)
                node['right_child'] = _get_node_index(tree['right_child'], tree_index)
                node['split_gain'] = tree['split_gain']
                node['threshold'] = tree['threshold']
                node['decision_type'] = tree['decision_type']
                node['missing_direction'] = 'left' if tree['default_left'] else 'right'
                node['missing_type'] = tree['missing_type']
                node['value'] = tree['internal_value']
                node['weight'] = tree['internal_weight']
                node['count'] = tree['internal_count']
            else:
                node['value'] = tree['leaf_value']
                if not _is_single_node_tree(tree):
                    node['weight'] = tree['leaf_weight']
                    node['count'] = tree['leaf_count']

            return node

        def tree_dict_to_node_list(tree, node_depth=1, tree_index=None,
                                   feature_names=None, parent_node=None):

            node = create_node_record(tree,
                                      node_depth=node_depth,
                                      tree_index=tree_index,
                                      feature_names=feature_names,
                                      parent_node=parent_node)

            res = [node]

            if _is_split_node(tree):
                # traverse the next level of the tree
                children = ['left_child', 'right_child']
                for child in children:
                    subtree_list = tree_dict_to_node_list(
                        tree[child],
                        node_depth=node_depth + 1,
                        tree_index=tree_index,
                        feature_names=feature_names,
                        parent_node=node['node_index'])
                    # In tree format, "subtree_list" is a list of node records (dicts),
                    # and we add node to the list.
                    res.extend(subtree_list)
            return res

        model_dict = self.dump_model()
        feature_names = model_dict['feature_names']
        model_list = []
        for tree in model_dict['tree_info']:
            model_list.extend(tree_dict_to_node_list(tree['tree_structure'],
                                                     tree_index=tree['tree_index'],
                                                     feature_names=feature_names))

2925
        return pd_DataFrame(model_list, columns=model_list[0].keys())
2926

wxchan's avatar
wxchan committed
2927
    def set_train_data_name(self, name):
2928
2929
2930
2931
        """Set the name to the training Dataset.

        Parameters
        ----------
2932
        name : str
Nikita Titov's avatar
Nikita Titov committed
2933
2934
2935
2936
2937
2938
            Name for the training Dataset.

        Returns
        -------
        self : Booster
            Booster with set training Dataset name.
2939
        """
2940
        self._train_data_name = name
Nikita Titov's avatar
Nikita Titov committed
2941
        return self
wxchan's avatar
wxchan committed
2942
2943

    def add_valid(self, data, name):
2944
        """Add validation data.
wxchan's avatar
wxchan committed
2945
2946
2947
2948

        Parameters
        ----------
        data : Dataset
2949
            Validation data.
2950
        name : str
2951
            Name of validation data.
Nikita Titov's avatar
Nikita Titov committed
2952
2953
2954
2955
2956

        Returns
        -------
        self : Booster
            Booster with set validation data.
wxchan's avatar
wxchan committed
2957
        """
Guolin Ke's avatar
Guolin Ke committed
2958
        if not isinstance(data, Dataset):
2959
            raise TypeError(f'Validation data should be Dataset instance, met {type(data).__name__}')
Guolin Ke's avatar
Guolin Ke committed
2960
        if data._predictor is not self.__init_predictor:
2961
2962
            raise LightGBMError("Add validation data failed, "
                                "you should use same predictor for these data")
wxchan's avatar
wxchan committed
2963
2964
        _safe_call(_LIB.LGBM_BoosterAddValidData(
            self.handle,
wxchan's avatar
wxchan committed
2965
            data.construct().handle))
wxchan's avatar
wxchan committed
2966
2967
2968
2969
2970
        self.valid_sets.append(data)
        self.name_valid_sets.append(name)
        self.__num_dataset += 1
        self.__inner_predict_buffer.append(None)
        self.__is_predicted_cur_iter.append(False)
Nikita Titov's avatar
Nikita Titov committed
2971
        return self
wxchan's avatar
wxchan committed
2972
2973

    def reset_parameter(self, params):
2974
        """Reset parameters of Booster.
wxchan's avatar
wxchan committed
2975
2976
2977
2978

        Parameters
        ----------
        params : dict
2979
            New parameters for Booster.
Nikita Titov's avatar
Nikita Titov committed
2980
2981
2982
2983
2984

        Returns
        -------
        self : Booster
            Booster with new parameters.
wxchan's avatar
wxchan committed
2985
2986
2987
2988
2989
2990
        """
        params_str = param_dict_to_str(params)
        if params_str:
            _safe_call(_LIB.LGBM_BoosterResetParameter(
                self.handle,
                c_str(params_str)))
Guolin Ke's avatar
Guolin Ke committed
2991
        self.params.update(params)
Nikita Titov's avatar
Nikita Titov committed
2992
        return self
wxchan's avatar
wxchan committed
2993
2994

    def update(self, train_set=None, fobj=None):
Nikita Titov's avatar
Nikita Titov committed
2995
        """Update Booster for one iteration.
2996

wxchan's avatar
wxchan committed
2997
2998
        Parameters
        ----------
2999
3000
3001
3002
        train_set : Dataset or None, optional (default=None)
            Training data.
            If None, last training data is used.
        fobj : callable or None, optional (default=None)
wxchan's avatar
wxchan committed
3003
            Customized objective function.
3004
3005
3006
3007
3008
            Should accept two parameters: preds, train_data,
            and return (grad, hess).

                preds : list or numpy 1-D array
                    The predicted values.
3009
3010
                    Predicted values are returned before any transformation,
                    e.g. they are raw margin instead of probability of positive class for binary task.
3011
3012
3013
                train_data : Dataset
                    The training dataset.
                grad : list or numpy 1-D array
3014
3015
                    The value of the first order derivative (gradient) of the loss
                    with respect to the elements of preds for each sample point.
3016
                hess : list or numpy 1-D array
3017
3018
                    The value of the second order derivative (Hessian) of the loss
                    with respect to the elements of preds for each sample point.
wxchan's avatar
wxchan committed
3019

3020
3021
            For multi-class task, the preds is group by class_id first, then group by row_id.
            If you want to get i-th row preds in j-th class, the access way is score[j * num_data + i]
3022
3023
            and you should group grad and hess in this way as well.

wxchan's avatar
wxchan committed
3024
3025
        Returns
        -------
3026
3027
        is_finished : bool
            Whether the update was successfully finished.
wxchan's avatar
wxchan committed
3028
        """
3029
        # need reset training data
3030
3031
3032
3033
3034
3035
        if train_set is None and self.train_set_version != self.train_set.version:
            train_set = self.train_set
            is_the_same_train_set = False
        else:
            is_the_same_train_set = train_set is self.train_set and self.train_set_version == train_set.version
        if train_set is not None and not is_the_same_train_set:
Guolin Ke's avatar
Guolin Ke committed
3036
            if not isinstance(train_set, Dataset):
3037
                raise TypeError(f'Training data should be Dataset instance, met {type(train_set).__name__}')
Guolin Ke's avatar
Guolin Ke committed
3038
            if train_set._predictor is not self.__init_predictor:
3039
3040
                raise LightGBMError("Replace training data failed, "
                                    "you should use same predictor for these data")
wxchan's avatar
wxchan committed
3041
3042
3043
            self.train_set = train_set
            _safe_call(_LIB.LGBM_BoosterResetTrainingData(
                self.handle,
wxchan's avatar
wxchan committed
3044
                self.train_set.construct().handle))
wxchan's avatar
wxchan committed
3045
            self.__inner_predict_buffer[0] = None
3046
            self.train_set_version = self.train_set.version
wxchan's avatar
wxchan committed
3047
3048
        is_finished = ctypes.c_int(0)
        if fobj is None:
3049
            if self.__set_objective_to_none:
3050
                raise LightGBMError('Cannot update due to null objective function.')
wxchan's avatar
wxchan committed
3051
3052
3053
            _safe_call(_LIB.LGBM_BoosterUpdateOneIter(
                self.handle,
                ctypes.byref(is_finished)))
3054
            self.__is_predicted_cur_iter = [False for _ in range(self.__num_dataset)]
wxchan's avatar
wxchan committed
3055
3056
            return is_finished.value == 1
        else:
3057
            if not self.__set_objective_to_none:
Nikita Titov's avatar
Nikita Titov committed
3058
                self.reset_parameter({"objective": "none"}).__set_objective_to_none = True
wxchan's avatar
wxchan committed
3059
3060
3061
3062
            grad, hess = fobj(self.__inner_predict(0), self.train_set)
            return self.__boost(grad, hess)

    def __boost(self, grad, hess):
3063
        """Boost Booster for one iteration with customized gradient statistics.
Nikita Titov's avatar
Nikita Titov committed
3064

Nikita Titov's avatar
Nikita Titov committed
3065
3066
        .. note::

3067
3068
            Score is returned before any transformation,
            e.g. it is raw margin instead of probability of positive class for binary task.
Nikita Titov's avatar
Nikita Titov committed
3069
3070
3071
            For multi-class task, the score is group by class_id first, then group by row_id.
            If you want to get i-th row score in j-th class, the access way is score[j * num_data + i]
            and you should group grad and hess in this way as well.
3072

wxchan's avatar
wxchan committed
3073
3074
        Parameters
        ----------
3075
        grad : list or numpy 1-D array
3076
3077
            The value of the first order derivative (gradient) of the loss
            with respect to the elements of score for each sample point.
3078
        hess : list or numpy 1-D array
3079
3080
            The value of the second order derivative (Hessian) of the loss
            with respect to the elements of score for each sample point.
wxchan's avatar
wxchan committed
3081
3082
3083

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
3084
3085
        is_finished : bool
            Whether the boost was successfully finished.
wxchan's avatar
wxchan committed
3086
        """
3087
3088
        grad = list_to_1d_numpy(grad, name='gradient')
        hess = list_to_1d_numpy(hess, name='hessian')
3089
3090
        assert grad.flags.c_contiguous
        assert hess.flags.c_contiguous
wxchan's avatar
wxchan committed
3091
        if len(grad) != len(hess):
3092
            raise ValueError(f"Lengths of gradient({len(grad)}) and hessian({len(hess)}) don't match")
wxchan's avatar
wxchan committed
3093
3094
3095
3096
3097
3098
        is_finished = ctypes.c_int(0)
        _safe_call(_LIB.LGBM_BoosterUpdateOneIterCustom(
            self.handle,
            grad.ctypes.data_as(ctypes.POINTER(ctypes.c_float)),
            hess.ctypes.data_as(ctypes.POINTER(ctypes.c_float)),
            ctypes.byref(is_finished)))
3099
        self.__is_predicted_cur_iter = [False for _ in range(self.__num_dataset)]
wxchan's avatar
wxchan committed
3100
3101
3102
        return is_finished.value == 1

    def rollback_one_iter(self):
Nikita Titov's avatar
Nikita Titov committed
3103
3104
3105
3106
3107
3108
3109
        """Rollback one iteration.

        Returns
        -------
        self : Booster
            Booster with rolled back one iteration.
        """
wxchan's avatar
wxchan committed
3110
3111
        _safe_call(_LIB.LGBM_BoosterRollbackOneIter(
            self.handle))
3112
        self.__is_predicted_cur_iter = [False for _ in range(self.__num_dataset)]
Nikita Titov's avatar
Nikita Titov committed
3113
        return self
wxchan's avatar
wxchan committed
3114
3115

    def current_iteration(self):
3116
3117
3118
3119
3120
3121
3122
        """Get the index of the current iteration.

        Returns
        -------
        cur_iter : int
            The index of the current iteration.
        """
Guolin Ke's avatar
Guolin Ke committed
3123
        out_cur_iter = ctypes.c_int(0)
wxchan's avatar
wxchan committed
3124
3125
3126
3127
3128
        _safe_call(_LIB.LGBM_BoosterGetCurrentIteration(
            self.handle,
            ctypes.byref(out_cur_iter)))
        return out_cur_iter.value

3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
    def num_model_per_iteration(self):
        """Get number of models per iteration.

        Returns
        -------
        model_per_iter : int
            The number of models per iteration.
        """
        model_per_iter = ctypes.c_int(0)
        _safe_call(_LIB.LGBM_BoosterNumModelPerIteration(
            self.handle,
            ctypes.byref(model_per_iter)))
        return model_per_iter.value

    def num_trees(self):
        """Get number of weak sub-models.

        Returns
        -------
        num_trees : int
            The number of weak sub-models.
        """
        num_trees = ctypes.c_int(0)
        _safe_call(_LIB.LGBM_BoosterNumberOfTotalModel(
            self.handle,
            ctypes.byref(num_trees)))
        return num_trees.value

3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
    def upper_bound(self):
        """Get upper bound value of a model.

        Returns
        -------
        upper_bound : double
            Upper bound value of the model.
        """
        ret = ctypes.c_double(0)
        _safe_call(_LIB.LGBM_BoosterGetUpperBoundValue(
            self.handle,
            ctypes.byref(ret)))
        return ret.value

    def lower_bound(self):
        """Get lower bound value of a model.

        Returns
        -------
        lower_bound : double
            Lower bound value of the model.
        """
        ret = ctypes.c_double(0)
        _safe_call(_LIB.LGBM_BoosterGetLowerBoundValue(
            self.handle,
            ctypes.byref(ret)))
        return ret.value

wxchan's avatar
wxchan committed
3185
    def eval(self, data, name, feval=None):
3186
        """Evaluate for data.
wxchan's avatar
wxchan committed
3187
3188
3189

        Parameters
        ----------
3190
3191
        data : Dataset
            Data for the evaluating.
3192
        name : str
3193
3194
            Name of the data.
        feval : callable or None, optional (default=None)
3195
            Customized evaluation function.
3196
            Should accept two parameters: preds, eval_data,
3197
            and return (eval_name, eval_result, is_higher_better) or list of such tuples.
3198
3199
3200

                preds : list or numpy 1-D array
                    The predicted values.
3201
3202
                    If ``fobj`` is specified, predicted values are returned before any transformation,
                    e.g. they are raw margin instead of probability of positive class for binary task in this case.
3203
3204
                eval_data : Dataset
                    The evaluation dataset.
3205
                eval_name : str
Andrew Ziem's avatar
Andrew Ziem committed
3206
                    The name of evaluation function (without whitespace).
3207
3208
3209
3210
3211
                eval_result : float
                    The eval result.
                is_higher_better : bool
                    Is eval result higher better, e.g. AUC is ``is_higher_better``.

3212
3213
            For multi-class task, the preds is group by class_id first, then group by row_id.
            If you want to get i-th row preds in j-th class, the access way is preds[j * num_data + i].
3214

wxchan's avatar
wxchan committed
3215
3216
        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
3217
        result : list
3218
            List with evaluation results.
wxchan's avatar
wxchan committed
3219
        """
Guolin Ke's avatar
Guolin Ke committed
3220
3221
        if not isinstance(data, Dataset):
            raise TypeError("Can only eval for Dataset instance")
wxchan's avatar
wxchan committed
3222
3223
3224
3225
        data_idx = -1
        if data is self.train_set:
            data_idx = 0
        else:
3226
            for i in range(len(self.valid_sets)):
wxchan's avatar
wxchan committed
3227
3228
3229
                if data is self.valid_sets[i]:
                    data_idx = i + 1
                    break
3230
        # need to push new valid data
wxchan's avatar
wxchan committed
3231
3232
3233
3234
3235
3236
3237
        if data_idx == -1:
            self.add_valid(data, name)
            data_idx = self.__num_dataset - 1

        return self.__inner_eval(name, data_idx, feval)

    def eval_train(self, feval=None):
3238
        """Evaluate for training data.
wxchan's avatar
wxchan committed
3239
3240
3241

        Parameters
        ----------
3242
        feval : callable or None, optional (default=None)
3243
            Customized evaluation function.
3244
3245
            Should accept two parameters: preds, train_data,
            and return (eval_name, eval_result, is_higher_better) or list of such tuples.
3246
3247
3248

                preds : list or numpy 1-D array
                    The predicted values.
3249
3250
                    If ``fobj`` is specified, predicted values are returned before any transformation,
                    e.g. they are raw margin instead of probability of positive class for binary task in this case.
3251
3252
                train_data : Dataset
                    The training dataset.
3253
                eval_name : str
Andrew Ziem's avatar
Andrew Ziem committed
3254
                    The name of evaluation function (without whitespace).
3255
3256
3257
3258
3259
                eval_result : float
                    The eval result.
                is_higher_better : bool
                    Is eval result higher better, e.g. AUC is ``is_higher_better``.

3260
3261
            For multi-class task, the preds is group by class_id first, then group by row_id.
            If you want to get i-th row preds in j-th class, the access way is preds[j * num_data + i].
wxchan's avatar
wxchan committed
3262
3263
3264

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
3265
        result : list
3266
            List with evaluation results.
wxchan's avatar
wxchan committed
3267
        """
3268
        return self.__inner_eval(self._train_data_name, 0, feval)
wxchan's avatar
wxchan committed
3269
3270

    def eval_valid(self, feval=None):
3271
        """Evaluate for validation data.
wxchan's avatar
wxchan committed
3272
3273
3274

        Parameters
        ----------
3275
        feval : callable or None, optional (default=None)
3276
            Customized evaluation function.
3277
            Should accept two parameters: preds, valid_data,
3278
            and return (eval_name, eval_result, is_higher_better) or list of such tuples.
3279
3280
3281

                preds : list or numpy 1-D array
                    The predicted values.
3282
3283
                    If ``fobj`` is specified, predicted values are returned before any transformation,
                    e.g. they are raw margin instead of probability of positive class for binary task in this case.
3284
3285
                valid_data : Dataset
                    The validation dataset.
3286
                eval_name : str
Andrew Ziem's avatar
Andrew Ziem committed
3287
                    The name of evaluation function (without whitespace).
3288
3289
3290
3291
3292
                eval_result : float
                    The eval result.
                is_higher_better : bool
                    Is eval result higher better, e.g. AUC is ``is_higher_better``.

3293
3294
            For multi-class task, the preds is group by class_id first, then group by row_id.
            If you want to get i-th row preds in j-th class, the access way is preds[j * num_data + i].
wxchan's avatar
wxchan committed
3295
3296
3297

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
3298
        result : list
3299
            List with evaluation results.
wxchan's avatar
wxchan committed
3300
        """
3301
        return [item for i in range(1, self.__num_dataset)
wxchan's avatar
wxchan committed
3302
                for item in self.__inner_eval(self.name_valid_sets[i - 1], i, feval)]
wxchan's avatar
wxchan committed
3303

3304
    def save_model(self, filename, num_iteration=None, start_iteration=0, importance_type='split'):
3305
        """Save Booster to file.
wxchan's avatar
wxchan committed
3306
3307
3308

        Parameters
        ----------
3309
        filename : str or pathlib.Path
3310
            Filename to save Booster.
3311
3312
3313
3314
        num_iteration : int or None, optional (default=None)
            Index of the iteration that should be saved.
            If None, if the best iteration exists, it is saved; otherwise, all iterations are saved.
            If <= 0, all iterations are saved.
Nikita Titov's avatar
Nikita Titov committed
3315
        start_iteration : int, optional (default=0)
3316
            Start index of the iteration that should be saved.
3317
        importance_type : str, optional (default="split")
3318
3319
3320
            What type of feature importance should be saved.
            If "split", result contains numbers of times the feature is used in a model.
            If "gain", result contains total gains of splits which use the feature.
Nikita Titov's avatar
Nikita Titov committed
3321
3322
3323
3324
3325

        Returns
        -------
        self : Booster
            Returns self.
wxchan's avatar
wxchan committed
3326
        """
3327
        if num_iteration is None:
3328
            num_iteration = self.best_iteration
3329
        importance_type_int = FEATURE_IMPORTANCE_TYPE_MAPPER[importance_type]
wxchan's avatar
wxchan committed
3330
3331
        _safe_call(_LIB.LGBM_BoosterSaveModel(
            self.handle,
3332
            ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
3333
            ctypes.c_int(num_iteration),
3334
            ctypes.c_int(importance_type_int),
3335
            c_str(str(filename))))
3336
        _dump_pandas_categorical(self.pandas_categorical, filename)
Nikita Titov's avatar
Nikita Titov committed
3337
        return self
wxchan's avatar
wxchan committed
3338

3339
    def shuffle_models(self, start_iteration=0, end_iteration=-1):
3340
        """Shuffle models.
Nikita Titov's avatar
Nikita Titov committed
3341

3342
3343
3344
        Parameters
        ----------
        start_iteration : int, optional (default=0)
3345
            The first iteration that will be shuffled.
3346
3347
        end_iteration : int, optional (default=-1)
            The last iteration that will be shuffled.
3348
            If <= 0, means the last available iteration.
3349

Nikita Titov's avatar
Nikita Titov committed
3350
3351
3352
3353
        Returns
        -------
        self : Booster
            Booster with shuffled models.
3354
        """
3355
3356
        _safe_call(_LIB.LGBM_BoosterShuffleModels(
            self.handle,
Guolin Ke's avatar
Guolin Ke committed
3357
3358
            ctypes.c_int(start_iteration),
            ctypes.c_int(end_iteration)))
Nikita Titov's avatar
Nikita Titov committed
3359
        return self
3360

3361
    def model_from_string(self, model_str, verbose='warn'):
3362
3363
3364
3365
        """Load Booster from a string.

        Parameters
        ----------
3366
        model_str : str
3367
            Model will be loaded from this string.
Nikita Titov's avatar
Nikita Titov committed
3368
3369
        verbose : bool, optional (default=True)
            Whether to print messages while loading model.
3370
3371
3372

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
3373
        self : Booster
3374
3375
            Loaded Booster object.
        """
3376
3377
3378
3379
        if self.handle is not None:
            _safe_call(_LIB.LGBM_BoosterFree(self.handle))
        self._free_buffer()
        self.handle = ctypes.c_void_p()
3380
3381
3382
3383
3384
3385
3386
3387
3388
        out_num_iterations = ctypes.c_int(0)
        _safe_call(_LIB.LGBM_BoosterLoadModelFromString(
            c_str(model_str),
            ctypes.byref(out_num_iterations),
            ctypes.byref(self.handle)))
        out_num_class = ctypes.c_int(0)
        _safe_call(_LIB.LGBM_BoosterGetNumClasses(
            self.handle,
            ctypes.byref(out_num_class)))
3389
3390
3391
3392
        if verbose in {'warn', '_silent_false'}:
            verbose = verbose == 'warn'
        else:
            _log_warning("'verbose' argument is deprecated and will be removed in a future release of LightGBM.")
3393
        if verbose:
3394
            _log_info(f'Finished loading model, total used {int(out_num_iterations.value)} iterations')
3395
        self.__num_class = out_num_class.value
3396
        self.pandas_categorical = _load_pandas_categorical(model_str=model_str)
3397
3398
        return self

3399
    def model_to_string(self, num_iteration=None, start_iteration=0, importance_type='split'):
3400
        """Save Booster to string.
3401

3402
3403
3404
3405
3406
3407
        Parameters
        ----------
        num_iteration : int or None, optional (default=None)
            Index of the iteration that should be saved.
            If None, if the best iteration exists, it is saved; otherwise, all iterations are saved.
            If <= 0, all iterations are saved.
Nikita Titov's avatar
Nikita Titov committed
3408
        start_iteration : int, optional (default=0)
3409
            Start index of the iteration that should be saved.
3410
        importance_type : str, optional (default="split")
3411
3412
3413
            What type of feature importance should be saved.
            If "split", result contains numbers of times the feature is used in a model.
            If "gain", result contains total gains of splits which use the feature.
3414
3415
3416

        Returns
        -------
3417
        str_repr : str
3418
3419
            String representation of Booster.
        """
3420
        if num_iteration is None:
3421
            num_iteration = self.best_iteration
3422
        importance_type_int = FEATURE_IMPORTANCE_TYPE_MAPPER[importance_type]
3423
        buffer_len = 1 << 20
3424
        tmp_out_len = ctypes.c_int64(0)
3425
3426
3427
3428
        string_buffer = ctypes.create_string_buffer(buffer_len)
        ptr_string_buffer = ctypes.c_char_p(*[ctypes.addressof(string_buffer)])
        _safe_call(_LIB.LGBM_BoosterSaveModelToString(
            self.handle,
3429
            ctypes.c_int(start_iteration),
3430
            ctypes.c_int(num_iteration),
3431
            ctypes.c_int(importance_type_int),
3432
            ctypes.c_int64(buffer_len),
3433
3434
3435
            ctypes.byref(tmp_out_len),
            ptr_string_buffer))
        actual_len = tmp_out_len.value
3436
        # if buffer length is not long enough, re-allocate a buffer
3437
3438
3439
3440
3441
        if actual_len > buffer_len:
            string_buffer = ctypes.create_string_buffer(actual_len)
            ptr_string_buffer = ctypes.c_char_p(*[ctypes.addressof(string_buffer)])
            _safe_call(_LIB.LGBM_BoosterSaveModelToString(
                self.handle,
3442
                ctypes.c_int(start_iteration),
3443
                ctypes.c_int(num_iteration),
3444
                ctypes.c_int(importance_type_int),
3445
                ctypes.c_int64(actual_len),
3446
3447
                ctypes.byref(tmp_out_len),
                ptr_string_buffer))
3448
        ret = string_buffer.value.decode('utf-8')
3449
3450
        ret += _dump_pandas_categorical(self.pandas_categorical)
        return ret
3451

3452
    def dump_model(self, num_iteration=None, start_iteration=0, importance_type='split', object_hook=None):
Nikita Titov's avatar
Nikita Titov committed
3453
        """Dump Booster to JSON format.
wxchan's avatar
wxchan committed
3454

3455
3456
        Parameters
        ----------
3457
3458
3459
3460
        num_iteration : int or None, optional (default=None)
            Index of the iteration that should be dumped.
            If None, if the best iteration exists, it is dumped; otherwise, all iterations are dumped.
            If <= 0, all iterations are dumped.
Nikita Titov's avatar
Nikita Titov committed
3461
        start_iteration : int, optional (default=0)
3462
            Start index of the iteration that should be dumped.
3463
        importance_type : str, optional (default="split")
3464
3465
3466
            What type of feature importance should be dumped.
            If "split", result contains numbers of times the feature is used in a model.
            If "gain", result contains total gains of splits which use the feature.
3467
3468
3469
3470
3471
3472
3473
3474
3475
        object_hook : callable or None, optional (default=None)
            If not None, ``object_hook`` is a function called while parsing the json
            string returned by the C API. It may be used to alter the json, to store
            specific values while building the json structure. It avoids
            walking through the structure again. It saves a significant amount
            of time if the number of trees is huge.
            Signature is ``def object_hook(node: dict) -> dict``.
            None is equivalent to ``lambda node: node``.
            See documentation of ``json.loads()`` for further details.
3476

wxchan's avatar
wxchan committed
3477
3478
        Returns
        -------
3479
        json_repr : dict
Nikita Titov's avatar
Nikita Titov committed
3480
            JSON format of Booster.
wxchan's avatar
wxchan committed
3481
        """
3482
        if num_iteration is None:
3483
            num_iteration = self.best_iteration
3484
        importance_type_int = FEATURE_IMPORTANCE_TYPE_MAPPER[importance_type]
wxchan's avatar
wxchan committed
3485
        buffer_len = 1 << 20
3486
        tmp_out_len = ctypes.c_int64(0)
wxchan's avatar
wxchan committed
3487
3488
3489
3490
        string_buffer = ctypes.create_string_buffer(buffer_len)
        ptr_string_buffer = ctypes.c_char_p(*[ctypes.addressof(string_buffer)])
        _safe_call(_LIB.LGBM_BoosterDumpModel(
            self.handle,
3491
            ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
3492
            ctypes.c_int(num_iteration),
3493
            ctypes.c_int(importance_type_int),
3494
            ctypes.c_int64(buffer_len),
wxchan's avatar
wxchan committed
3495
            ctypes.byref(tmp_out_len),
Guolin Ke's avatar
Guolin Ke committed
3496
            ptr_string_buffer))
wxchan's avatar
wxchan committed
3497
        actual_len = tmp_out_len.value
3498
        # if buffer length is not long enough, reallocate a buffer
wxchan's avatar
wxchan committed
3499
3500
3501
3502
3503
        if actual_len > buffer_len:
            string_buffer = ctypes.create_string_buffer(actual_len)
            ptr_string_buffer = ctypes.c_char_p(*[ctypes.addressof(string_buffer)])
            _safe_call(_LIB.LGBM_BoosterDumpModel(
                self.handle,
3504
                ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
3505
                ctypes.c_int(num_iteration),
3506
                ctypes.c_int(importance_type_int),
3507
                ctypes.c_int64(actual_len),
wxchan's avatar
wxchan committed
3508
                ctypes.byref(tmp_out_len),
Guolin Ke's avatar
Guolin Ke committed
3509
                ptr_string_buffer))
3510
        ret = json.loads(string_buffer.value.decode('utf-8'), object_hook=object_hook)
3511
3512
3513
        ret['pandas_categorical'] = json.loads(json.dumps(self.pandas_categorical,
                                                          default=json_default_with_numpy))
        return ret
wxchan's avatar
wxchan committed
3514

3515
    def predict(self, data, start_iteration=0, num_iteration=None,
3516
                raw_score=False, pred_leaf=False, pred_contrib=False,
3517
                data_has_header=False, is_reshape=True, **kwargs):
3518
        """Make a prediction.
wxchan's avatar
wxchan committed
3519
3520
3521

        Parameters
        ----------
3522
        data : str, pathlib.Path, numpy array, pandas DataFrame, H2O DataTable's Frame or scipy.sparse
3523
            Data source for prediction.
3524
            If str or pathlib.Path, it represents the path to a text file (CSV, TSV, or LibSVM).
3525
        start_iteration : int, optional (default=0)
3526
            Start index of the iteration to predict.
3527
            If <= 0, starts from the first iteration.
3528
        num_iteration : int or None, optional (default=None)
3529
3530
3531
3532
            Total number of iterations used in the prediction.
            If None, if the best iteration exists and start_iteration <= 0, the best iteration is used;
            otherwise, all iterations from ``start_iteration`` are used (no limits).
            If <= 0, all iterations from ``start_iteration`` are used (no limits).
3533
3534
3535
3536
        raw_score : bool, optional (default=False)
            Whether to predict raw scores.
        pred_leaf : bool, optional (default=False)
            Whether to predict leaf index.
3537
3538
        pred_contrib : bool, optional (default=False)
            Whether to predict feature contributions.
3539

Nikita Titov's avatar
Nikita Titov committed
3540
3541
3542
3543
3544
3545
3546
            .. note::

                If you want to get more explanations for your model's predictions using SHAP values,
                like SHAP interaction values,
                you can install the shap package (https://github.com/slundberg/shap).
                Note that unlike the shap package, with ``pred_contrib`` we return a matrix with an extra
                column, where the last column is the expected value.
3547

3548
3549
        data_has_header : bool, optional (default=False)
            Whether the data has header.
3550
            Used only if data is str.
3551
3552
        is_reshape : bool, optional (default=True)
            If True, result is reshaped to [nrow, ncol].
3553
3554
        **kwargs
            Other parameters for the prediction.
wxchan's avatar
wxchan committed
3555
3556
3557

        Returns
        -------
3558
        result : numpy array, scipy.sparse or list of scipy.sparse
3559
            Prediction result.
3560
            Can be sparse or a list of sparse objects (each element represents predictions for one class) for feature contributions (when ``pred_contrib=True``).
wxchan's avatar
wxchan committed
3561
        """
3562
        predictor = self._to_predictor(deepcopy(kwargs))
3563
        if num_iteration is None:
3564
            if start_iteration <= 0:
3565
3566
3567
3568
                num_iteration = self.best_iteration
            else:
                num_iteration = -1
        return predictor.predict(data, start_iteration, num_iteration,
3569
3570
                                 raw_score, pred_leaf, pred_contrib,
                                 data_has_header, is_reshape)
wxchan's avatar
wxchan committed
3571

3572
    def refit(self, data, label, decay_rate=0.9, **kwargs):
Guolin Ke's avatar
Guolin Ke committed
3573
3574
3575
3576
        """Refit the existing Booster by new data.

        Parameters
        ----------
3577
        data : str, pathlib.Path, numpy array, pandas DataFrame, H2O DataTable's Frame or scipy.sparse
Guolin Ke's avatar
Guolin Ke committed
3578
            Data source for refit.
3579
            If str or pathlib.Path, it represents the path to a text file (CSV, TSV, or LibSVM).
3580
        label : list, numpy 1-D array or pandas Series / one-column DataFrame
Guolin Ke's avatar
Guolin Ke committed
3581
3582
            Label for refit.
        decay_rate : float, optional (default=0.9)
3583
3584
            Decay rate of refit,
            will use ``leaf_output = decay_rate * old_leaf_output + (1.0 - decay_rate) * new_leaf_output`` to refit trees.
3585
3586
        **kwargs
            Other parameters for refit.
3587
            These parameters will be passed to ``predict`` method.
Guolin Ke's avatar
Guolin Ke committed
3588
3589
3590
3591
3592
3593

        Returns
        -------
        result : Booster
            Refitted Booster.
        """
3594
3595
        if self.__set_objective_to_none:
            raise LightGBMError('Cannot refit due to null objective function.')
3596
        predictor = self._to_predictor(deepcopy(kwargs))
3597
        leaf_preds = predictor.predict(data, -1, pred_leaf=True)
3598
        nrow, ncol = leaf_preds.shape
3599
3600
3601
3602
        out_is_linear = ctypes.c_bool(False)
        _safe_call(_LIB.LGBM_BoosterGetLinear(
            self.handle,
            ctypes.byref(out_is_linear)))
Nikita Titov's avatar
Nikita Titov committed
3603
3604
3605
3606
3607
        new_params = _choose_param_value(
            main_param_name="linear_tree",
            params=self.params,
            default_value=None
        )
3608
3609
        new_params["linear_tree"] = out_is_linear.value
        train_set = Dataset(data, label, silent=True, params=new_params)
3610
        new_params['refit_decay_rate'] = decay_rate
3611
        new_booster = Booster(new_params, train_set)
Guolin Ke's avatar
Guolin Ke committed
3612
3613
3614
3615
3616
        # Copy models
        _safe_call(_LIB.LGBM_BoosterMerge(
            new_booster.handle,
            predictor.handle))
        leaf_preds = leaf_preds.reshape(-1)
3617
        ptr_data, _, _ = c_int_array(leaf_preds)
Guolin Ke's avatar
Guolin Ke committed
3618
3619
3620
        _safe_call(_LIB.LGBM_BoosterRefit(
            new_booster.handle,
            ptr_data,
3621
3622
            ctypes.c_int32(nrow),
            ctypes.c_int32(ncol)))
3623
3624
        new_booster.network = self.network
        new_booster.__attr = self.__attr.copy()
Guolin Ke's avatar
Guolin Ke committed
3625
3626
        return new_booster

3627
    def get_leaf_output(self, tree_id, leaf_id):
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
        """Get the output of a leaf.

        Parameters
        ----------
        tree_id : int
            The index of the tree.
        leaf_id : int
            The index of the leaf in the tree.

        Returns
        -------
        result : float
            The output of the leaf.
        """
3642
3643
3644
3645
3646
3647
3648
3649
        ret = ctypes.c_double(0)
        _safe_call(_LIB.LGBM_BoosterGetLeafValue(
            self.handle,
            ctypes.c_int(tree_id),
            ctypes.c_int(leaf_id),
            ctypes.byref(ret)))
        return ret.value

3650
    def _to_predictor(self, pred_parameter=None):
3651
        """Convert to predictor."""
3652
        predictor = _InnerPredictor(booster_handle=self.handle, pred_parameter=pred_parameter)
3653
        predictor.pandas_categorical = self.pandas_categorical
wxchan's avatar
wxchan committed
3654
3655
        return predictor

3656
    def num_feature(self):
3657
3658
3659
3660
3661
3662
3663
        """Get number of features.

        Returns
        -------
        num_feature : int
            The number of features.
        """
3664
3665
3666
3667
3668
3669
        out_num_feature = ctypes.c_int(0)
        _safe_call(_LIB.LGBM_BoosterGetNumFeature(
            self.handle,
            ctypes.byref(out_num_feature)))
        return out_num_feature.value

wxchan's avatar
wxchan committed
3670
    def feature_name(self):
3671
        """Get names of features.
wxchan's avatar
wxchan committed
3672
3673
3674

        Returns
        -------
3675
3676
        result : list
            List with names of features.
wxchan's avatar
wxchan committed
3677
        """
3678
        num_feature = self.num_feature()
3679
        # Get name of features
wxchan's avatar
wxchan committed
3680
        tmp_out_len = ctypes.c_int(0)
3681
3682
        reserved_string_buffer_size = 255
        required_string_buffer_size = ctypes.c_size_t(0)
3683
        string_buffers = [ctypes.create_string_buffer(reserved_string_buffer_size) for _ in range(num_feature)]
wxchan's avatar
wxchan committed
3684
3685
3686
        ptr_string_buffers = (ctypes.c_char_p * num_feature)(*map(ctypes.addressof, string_buffers))
        _safe_call(_LIB.LGBM_BoosterGetFeatureNames(
            self.handle,
3687
            ctypes.c_int(num_feature),
wxchan's avatar
wxchan committed
3688
            ctypes.byref(tmp_out_len),
3689
            ctypes.c_size_t(reserved_string_buffer_size),
3690
            ctypes.byref(required_string_buffer_size),
wxchan's avatar
wxchan committed
3691
3692
3693
            ptr_string_buffers))
        if num_feature != tmp_out_len.value:
            raise ValueError("Length of feature names doesn't equal with num_feature")
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
        actual_string_buffer_size = required_string_buffer_size.value
        # if buffer length is not long enough, reallocate buffers
        if reserved_string_buffer_size < actual_string_buffer_size:
            string_buffers = [ctypes.create_string_buffer(actual_string_buffer_size) for _ in range(num_feature)]
            ptr_string_buffers = (ctypes.c_char_p * num_feature)(*map(ctypes.addressof, string_buffers))
            _safe_call(_LIB.LGBM_BoosterGetFeatureNames(
                self.handle,
                ctypes.c_int(num_feature),
                ctypes.byref(tmp_out_len),
                ctypes.c_size_t(actual_string_buffer_size),
                ctypes.byref(required_string_buffer_size),
                ptr_string_buffers))
3706
        return [string_buffers[i].value.decode('utf-8') for i in range(num_feature)]
wxchan's avatar
wxchan committed
3707

3708
    def feature_importance(self, importance_type='split', iteration=None):
3709
        """Get feature importances.
3710

3711
3712
        Parameters
        ----------
3713
        importance_type : str, optional (default="split")
3714
3715
3716
            How the importance is calculated.
            If "split", result contains numbers of times the feature is used in a model.
            If "gain", result contains total gains of splits which use the feature.
3717
3718
3719
3720
        iteration : int or None, optional (default=None)
            Limit number of iterations in the feature importance calculation.
            If None, if the best iteration exists, it is used; otherwise, all trees are used.
            If <= 0, all trees are used (no limits).
3721

3722
3723
        Returns
        -------
3724
3725
        result : numpy array
            Array with feature importances.
3726
        """
3727
3728
        if iteration is None:
            iteration = self.best_iteration
3729
        importance_type_int = FEATURE_IMPORTANCE_TYPE_MAPPER[importance_type]
3730
        result = np.empty(self.num_feature(), dtype=np.float64)
3731
3732
3733
3734
3735
3736
        _safe_call(_LIB.LGBM_BoosterFeatureImportance(
            self.handle,
            ctypes.c_int(iteration),
            ctypes.c_int(importance_type_int),
            result.ctypes.data_as(ctypes.POINTER(ctypes.c_double))))
        if importance_type_int == 0:
3737
            return result.astype(np.int32)
3738
3739
        else:
            return result
3740

3741
3742
3743
3744
3745
    def get_split_value_histogram(self, feature, bins=None, xgboost_style=False):
        """Get split value histogram for the specified feature.

        Parameters
        ----------
3746
        feature : int or str
3747
3748
            The feature name or index the histogram is calculated for.
            If int, interpreted as index.
3749
            If str, interpreted as name.
3750

Nikita Titov's avatar
Nikita Titov committed
3751
3752
3753
            .. warning::

                Categorical features are not supported.
3754

3755
        bins : int, str or None, optional (default=None)
3756
3757
3758
            The maximum number of bins.
            If None, or int and > number of unique split values and ``xgboost_style=True``,
            the number of bins equals number of unique split values.
3759
            If str, it should be one from the list of the supported values by ``numpy.histogram()`` function.
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
        xgboost_style : bool, optional (default=False)
            Whether the returned result should be in the same form as it is in XGBoost.
            If False, the returned value is tuple of 2 numpy arrays as it is in ``numpy.histogram()`` function.
            If True, the returned value is matrix, in which the first column is the right edges of non-empty bins
            and the second one is the histogram values.

        Returns
        -------
        result_tuple : tuple of 2 numpy arrays
            If ``xgboost_style=False``, the values of the histogram of used splitting values for the specified feature
            and the bin edges.
        result_array_like : numpy array or pandas DataFrame (if pandas is installed)
            If ``xgboost_style=True``, the histogram of used splitting values for the specified feature.
        """
        def add(root):
            """Recursively add thresholds."""
            if 'split_index' in root:  # non-leaf
3777
                if feature_names is not None and isinstance(feature, str):
3778
3779
3780
3781
                    split_feature = feature_names[root['split_feature']]
                else:
                    split_feature = root['split_feature']
                if split_feature == feature:
3782
                    if isinstance(root['threshold'], str):
3783
3784
3785
                        raise LightGBMError('Cannot compute split value histogram for the categorical feature')
                    else:
                        values.append(root['threshold'])
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
                add(root['left_child'])
                add(root['right_child'])

        model = self.dump_model()
        feature_names = model.get('feature_names')
        tree_infos = model['tree_info']
        values = []
        for tree_info in tree_infos:
            add(tree_info['tree_structure'])

3796
        if bins is None or isinstance(bins, int) and xgboost_style:
3797
3798
3799
3800
3801
3802
3803
            n_unique = len(np.unique(values))
            bins = max(min(n_unique, bins) if bins is not None else n_unique, 1)
        hist, bin_edges = np.histogram(values, bins=bins)
        if xgboost_style:
            ret = np.column_stack((bin_edges[1:], hist))
            ret = ret[ret[:, 1] > 0]
            if PANDAS_INSTALLED:
3804
                return pd_DataFrame(ret, columns=['SplitValue', 'Count'])
3805
3806
3807
3808
3809
            else:
                return ret
        else:
            return hist, bin_edges

wxchan's avatar
wxchan committed
3810
    def __inner_eval(self, data_name, data_idx, feval=None):
3811
        """Evaluate training or validation data."""
wxchan's avatar
wxchan committed
3812
        if data_idx >= self.__num_dataset:
3813
            raise ValueError("Data_idx should be smaller than number of dataset")
wxchan's avatar
wxchan committed
3814
3815
3816
        self.__get_eval_info()
        ret = []
        if self.__num_inner_eval > 0:
3817
            result = np.empty(self.__num_inner_eval, dtype=np.float64)
Guolin Ke's avatar
Guolin Ke committed
3818
            tmp_out_len = ctypes.c_int(0)
wxchan's avatar
wxchan committed
3819
3820
            _safe_call(_LIB.LGBM_BoosterGetEval(
                self.handle,
Guolin Ke's avatar
Guolin Ke committed
3821
                ctypes.c_int(data_idx),
wxchan's avatar
wxchan committed
3822
                ctypes.byref(tmp_out_len),
Guolin Ke's avatar
Guolin Ke committed
3823
                result.ctypes.data_as(ctypes.POINTER(ctypes.c_double))))
wxchan's avatar
wxchan committed
3824
            if tmp_out_len.value != self.__num_inner_eval:
3825
                raise ValueError("Wrong length of eval results")
3826
            for i in range(self.__num_inner_eval):
3827
3828
                ret.append((data_name, self.__name_inner_eval[i],
                            result[i], self.__higher_better_inner_eval[i]))
3829
3830
        if callable(feval):
            feval = [feval]
wxchan's avatar
wxchan committed
3831
3832
3833
3834
3835
        if feval is not None:
            if data_idx == 0:
                cur_data = self.train_set
            else:
                cur_data = self.valid_sets[data_idx - 1]
3836
3837
3838
3839
3840
3841
3842
3843
3844
            for eval_function in feval:
                if eval_function is None:
                    continue
                feval_ret = eval_function(self.__inner_predict(data_idx), cur_data)
                if isinstance(feval_ret, list):
                    for eval_name, val, is_higher_better in feval_ret:
                        ret.append((data_name, eval_name, val, is_higher_better))
                else:
                    eval_name, val, is_higher_better = feval_ret
wxchan's avatar
wxchan committed
3845
3846
3847
3848
                    ret.append((data_name, eval_name, val, is_higher_better))
        return ret

    def __inner_predict(self, data_idx):
3849
        """Predict for training and validation dataset."""
wxchan's avatar
wxchan committed
3850
        if data_idx >= self.__num_dataset:
3851
            raise ValueError("Data_idx should be smaller than number of dataset")
wxchan's avatar
wxchan committed
3852
3853
3854
3855
3856
        if self.__inner_predict_buffer[data_idx] is None:
            if data_idx == 0:
                n_preds = self.train_set.num_data() * self.__num_class
            else:
                n_preds = self.valid_sets[data_idx - 1].num_data() * self.__num_class
3857
            self.__inner_predict_buffer[data_idx] = np.empty(n_preds, dtype=np.float64)
3858
        # avoid to predict many time in one iteration
wxchan's avatar
wxchan committed
3859
3860
        if not self.__is_predicted_cur_iter[data_idx]:
            tmp_out_len = ctypes.c_int64(0)
Guolin Ke's avatar
Guolin Ke committed
3861
            data_ptr = self.__inner_predict_buffer[data_idx].ctypes.data_as(ctypes.POINTER(ctypes.c_double))
wxchan's avatar
wxchan committed
3862
3863
            _safe_call(_LIB.LGBM_BoosterGetPredict(
                self.handle,
Guolin Ke's avatar
Guolin Ke committed
3864
                ctypes.c_int(data_idx),
wxchan's avatar
wxchan committed
3865
3866
3867
                ctypes.byref(tmp_out_len),
                data_ptr))
            if tmp_out_len.value != len(self.__inner_predict_buffer[data_idx]):
3868
                raise ValueError(f"Wrong length of predict results for data {data_idx}")
wxchan's avatar
wxchan committed
3869
3870
3871
3872
            self.__is_predicted_cur_iter[data_idx] = True
        return self.__inner_predict_buffer[data_idx]

    def __get_eval_info(self):
3873
        """Get inner evaluation count and names."""
wxchan's avatar
wxchan committed
3874
3875
        if self.__need_reload_eval_info:
            self.__need_reload_eval_info = False
Guolin Ke's avatar
Guolin Ke committed
3876
            out_num_eval = ctypes.c_int(0)
3877
            # Get num of inner evals
wxchan's avatar
wxchan committed
3878
3879
3880
3881
3882
            _safe_call(_LIB.LGBM_BoosterGetEvalCounts(
                self.handle,
                ctypes.byref(out_num_eval)))
            self.__num_inner_eval = out_num_eval.value
            if self.__num_inner_eval > 0:
3883
                # Get name of eval metrics
Guolin Ke's avatar
Guolin Ke committed
3884
                tmp_out_len = ctypes.c_int(0)
3885
3886
3887
                reserved_string_buffer_size = 255
                required_string_buffer_size = ctypes.c_size_t(0)
                string_buffers = [
3888
                    ctypes.create_string_buffer(reserved_string_buffer_size) for _ in range(self.__num_inner_eval)
3889
                ]
wxchan's avatar
wxchan committed
3890
                ptr_string_buffers = (ctypes.c_char_p * self.__num_inner_eval)(*map(ctypes.addressof, string_buffers))
wxchan's avatar
wxchan committed
3891
3892
                _safe_call(_LIB.LGBM_BoosterGetEvalNames(
                    self.handle,
3893
                    ctypes.c_int(self.__num_inner_eval),
wxchan's avatar
wxchan committed
3894
                    ctypes.byref(tmp_out_len),
3895
                    ctypes.c_size_t(reserved_string_buffer_size),
3896
                    ctypes.byref(required_string_buffer_size),
wxchan's avatar
wxchan committed
3897
3898
                    ptr_string_buffers))
                if self.__num_inner_eval != tmp_out_len.value:
3899
                    raise ValueError("Length of eval names doesn't equal with num_evals")
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
                actual_string_buffer_size = required_string_buffer_size.value
                # if buffer length is not long enough, reallocate buffers
                if reserved_string_buffer_size < actual_string_buffer_size:
                    string_buffers = [
                        ctypes.create_string_buffer(actual_string_buffer_size) for _ in range(self.__num_inner_eval)
                    ]
                    ptr_string_buffers = (ctypes.c_char_p * self.__num_inner_eval)(*map(ctypes.addressof, string_buffers))
                    _safe_call(_LIB.LGBM_BoosterGetEvalNames(
                        self.handle,
                        ctypes.c_int(self.__num_inner_eval),
                        ctypes.byref(tmp_out_len),
                        ctypes.c_size_t(actual_string_buffer_size),
                        ctypes.byref(required_string_buffer_size),
                        ptr_string_buffers))
                self.__name_inner_eval = [
                    string_buffers[i].value.decode('utf-8') for i in range(self.__num_inner_eval)
                ]
                self.__higher_better_inner_eval = [
                    name.startswith(('auc', 'ndcg@', 'map@', 'average_precision')) for name in self.__name_inner_eval
                ]
3920

wxchan's avatar
wxchan committed
3921
    def attr(self, key):
3922
        """Get attribute string from the Booster.
wxchan's avatar
wxchan committed
3923
3924
3925

        Parameters
        ----------
3926
        key : str
3927
            The name of the attribute.
wxchan's avatar
wxchan committed
3928
3929
3930

        Returns
        -------
3931
        value : str or None
3932
            The attribute value.
Nikita Titov's avatar
Nikita Titov committed
3933
            Returns None if attribute does not exist.
wxchan's avatar
wxchan committed
3934
        """
3935
        return self.__attr.get(key, None)
wxchan's avatar
wxchan committed
3936
3937

    def set_attr(self, **kwargs):
3938
        """Set attributes to the Booster.
wxchan's avatar
wxchan committed
3939
3940
3941
3942

        Parameters
        ----------
        **kwargs
3943
3944
            The attributes to set.
            Setting a value to None deletes an attribute.
Nikita Titov's avatar
Nikita Titov committed
3945
3946
3947
3948

        Returns
        -------
        self : Booster
3949
            Booster with set attributes.
wxchan's avatar
wxchan committed
3950
3951
3952
        """
        for key, value in kwargs.items():
            if value is not None:
3953
                if not isinstance(value, str):
Nikita Titov's avatar
Nikita Titov committed
3954
                    raise ValueError("Only string values are accepted")
wxchan's avatar
wxchan committed
3955
3956
3957
                self.__attr[key] = value
            else:
                self.__attr.pop(key, None)
Nikita Titov's avatar
Nikita Titov committed
3958
        return self