basic.py 141 KB
Newer Older
wxchan's avatar
wxchan committed
1
# coding: utf-8
2
"""Wrapper for C API of LightGBM."""
3
import copy
wxchan's avatar
wxchan committed
4
import ctypes
5
import json
6
import os
wxchan's avatar
wxchan committed
7
import warnings
wxchan's avatar
wxchan committed
8
from tempfile import NamedTemporaryFile
9
from collections import OrderedDict
wxchan's avatar
wxchan committed
10
11
12
13

import numpy as np
import scipy.sparse

14
from .compat import PANDAS_INSTALLED, DataFrame, Series, is_dtype_sparse, DataTable
wxchan's avatar
wxchan committed
15
16
from .libpath import find_lib_path

wxchan's avatar
wxchan committed
17

18
19
def _log_callback(msg):
    """Redirect logs from native library into Python console."""
20
    print("{0:s}".format(msg.decode('utf-8')), end='')
21
22


wxchan's avatar
wxchan committed
23
def _load_lib():
24
    """Load LightGBM library."""
wxchan's avatar
wxchan committed
25
26
    lib_path = find_lib_path()
    if len(lib_path) == 0:
27
        return None
wxchan's avatar
wxchan committed
28
29
    lib = ctypes.cdll.LoadLibrary(lib_path[0])
    lib.LGBM_GetLastError.restype = ctypes.c_char_p
30
31
32
    callback = ctypes.CFUNCTYPE(None, ctypes.c_char_p)
    lib.callback = callback(_log_callback)
    if lib.LGBM_RegisterLogCallback(lib.callback) != 0:
33
        raise LightGBMError(lib.LGBM_GetLastError().decode('utf-8'))
wxchan's avatar
wxchan committed
34
35
    return lib

wxchan's avatar
wxchan committed
36

wxchan's avatar
wxchan committed
37
38
_LIB = _load_lib()

wxchan's avatar
wxchan committed
39

40
41
42
NUMERIC_TYPES = (int, float, bool)


wxchan's avatar
wxchan committed
43
def _safe_call(ret):
44
45
    """Check the return value from C API call.

wxchan's avatar
wxchan committed
46
47
48
    Parameters
    ----------
    ret : int
49
        The return value from C API calls.
wxchan's avatar
wxchan committed
50
51
    """
    if ret != 0:
52
        raise LightGBMError(_LIB.LGBM_GetLastError().decode('utf-8'))
wxchan's avatar
wxchan committed
53

wxchan's avatar
wxchan committed
54

wxchan's avatar
wxchan committed
55
def is_numeric(obj):
56
    """Check whether object is a number or not, include numpy number, etc."""
wxchan's avatar
wxchan committed
57
58
59
    try:
        float(obj)
        return True
wxchan's avatar
wxchan committed
60
61
62
    except (TypeError, ValueError):
        # TypeError: obj is not a string or a number
        # ValueError: invalid literal
wxchan's avatar
wxchan committed
63
64
        return False

wxchan's avatar
wxchan committed
65

wxchan's avatar
wxchan committed
66
def is_numpy_1d_array(data):
67
    """Check whether data is a numpy 1-D array."""
68
    return isinstance(data, np.ndarray) and len(data.shape) == 1
wxchan's avatar
wxchan committed
69

wxchan's avatar
wxchan committed
70

wxchan's avatar
wxchan committed
71
def is_1d_list(data):
72
73
    """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
74

wxchan's avatar
wxchan committed
75

76
def list_to_1d_numpy(data, dtype=np.float32, name='list'):
77
    """Convert data to numpy 1-D array."""
wxchan's avatar
wxchan committed
78
79
80
81
82
83
84
    if is_numpy_1d_array(data):
        if data.dtype == dtype:
            return data
        else:
            return data.astype(dtype=dtype, copy=False)
    elif is_1d_list(data):
        return np.array(data, dtype=dtype, copy=False)
85
    elif isinstance(data, Series):
86
87
        if _get_bad_pandas_dtypes([data.dtypes]):
            raise ValueError('Series.dtypes must be int, float or bool')
88
        return np.array(data, dtype=dtype, copy=False)  # SparseArray should be supported as well
wxchan's avatar
wxchan committed
89
    else:
90
91
        raise TypeError("Wrong type({0}) for {1}.\n"
                        "It should be list, numpy 1-D array or pandas Series".format(type(data).__name__, name))
wxchan's avatar
wxchan committed
92

wxchan's avatar
wxchan committed
93

wxchan's avatar
wxchan committed
94
def cfloat32_array_to_numpy(cptr, length):
95
    """Convert a ctypes float pointer array to a numpy array."""
wxchan's avatar
wxchan committed
96
    if isinstance(cptr, ctypes.POINTER(ctypes.c_float)):
97
        return np.fromiter(cptr, dtype=np.float32, count=length)
wxchan's avatar
wxchan committed
98
    else:
99
        raise RuntimeError('Expected float pointer')
wxchan's avatar
wxchan committed
100

Guolin Ke's avatar
Guolin Ke committed
101

Guolin Ke's avatar
Guolin Ke committed
102
def cfloat64_array_to_numpy(cptr, length):
103
    """Convert a ctypes double pointer array to a numpy array."""
Guolin Ke's avatar
Guolin Ke committed
104
105
106
107
108
    if isinstance(cptr, ctypes.POINTER(ctypes.c_double)):
        return np.fromiter(cptr, dtype=np.float64, count=length)
    else:
        raise RuntimeError('Expected double pointer')

wxchan's avatar
wxchan committed
109

wxchan's avatar
wxchan committed
110
def cint32_array_to_numpy(cptr, length):
111
    """Convert a ctypes int pointer array to a numpy array."""
wxchan's avatar
wxchan committed
112
    if isinstance(cptr, ctypes.POINTER(ctypes.c_int32)):
113
        return np.fromiter(cptr, dtype=np.int32, count=length)
wxchan's avatar
wxchan committed
114
    else:
115
116
117
118
119
120
121
122
123
        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)):
        return np.fromiter(cptr, dtype=np.int64, count=length)
    else:
        raise RuntimeError('Expected int64 pointer')
wxchan's avatar
wxchan committed
124

wxchan's avatar
wxchan committed
125

wxchan's avatar
wxchan committed
126
def c_str(string):
127
    """Convert a Python string to C string."""
wxchan's avatar
wxchan committed
128
129
    return ctypes.c_char_p(string.encode('utf-8'))

wxchan's avatar
wxchan committed
130

wxchan's avatar
wxchan committed
131
def c_array(ctype, values):
132
    """Convert a Python array to C array."""
wxchan's avatar
wxchan committed
133
134
    return (ctype * len(values))(*values)

wxchan's avatar
wxchan committed
135

136
137
138
139
140
141
142
143
144
145
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
146
def param_dict_to_str(data):
147
    """Convert Python dictionary to string, which is passed to C API."""
148
    if data is None or not data:
wxchan's avatar
wxchan committed
149
150
151
        return ""
    pairs = []
    for key, val in data.items():
152
        if isinstance(val, (list, tuple, set)) or is_numpy_1d_array(val):
153
154
155
156
157
158
            def to_string(x):
                if isinstance(x, list):
                    return "[{}]".format(','.join(map(str, x)))
                else:
                    return str(x)
            pairs.append(str(key) + '=' + ','.join(map(to_string, val)))
159
        elif isinstance(val, (str, NUMERIC_TYPES)) or is_numeric(val):
wxchan's avatar
wxchan committed
160
            pairs.append(str(key) + '=' + str(val))
161
        elif val is not None:
162
            raise TypeError('Unknown type of parameter:%s, got:%s'
wxchan's avatar
wxchan committed
163
164
                            % (key, type(val).__name__))
    return ' '.join(pairs)
165

wxchan's avatar
wxchan committed
166

167
class _TempFile:
168
169
170
171
    def __enter__(self):
        with NamedTemporaryFile(prefix="lightgbm_tmp_", delete=True) as f:
            self.name = f.name
        return self
wxchan's avatar
wxchan committed
172

173
174
175
    def __exit__(self, exc_type, exc_val, exc_tb):
        if os.path.isfile(self.name):
            os.remove(self.name)
wxchan's avatar
wxchan committed
176

177
178
179
180
    def readlines(self):
        with open(self.name, "r+") as f:
            ret = f.readlines()
        return ret
wxchan's avatar
wxchan committed
181

182
183
    def writelines(self, lines):
        with open(self.name, "w+") as f:
184
            f.writelines(lines)
185

wxchan's avatar
wxchan committed
186

187
class LightGBMError(Exception):
188
189
    """Error thrown by LightGBM."""

190
191
192
    pass


193
194
195
196
197
198
199
200
# DeprecationWarning is not shown by default, so let's create our own with higher level
class LGBMDeprecationWarning(UserWarning):
    """Custom deprecation warning."""

    pass


class _ConfigAliases:
201
202
203
    aliases = {"bin_construct_sample_cnt": {"bin_construct_sample_cnt",
                                            "subsample_for_bin"},
               "boosting": {"boosting",
204
205
206
207
208
209
                            "boosting_type",
                            "boost"},
               "categorical_feature": {"categorical_feature",
                                       "cat_feature",
                                       "categorical_column",
                                       "cat_column"},
210
211
               "data_random_seed": {"data_random_seed",
                                    "data_seed"},
212
213
214
215
               "early_stopping_round": {"early_stopping_round",
                                        "early_stopping_rounds",
                                        "early_stopping",
                                        "n_iter_no_change"},
216
217
218
               "enable_bundle": {"enable_bundle",
                                 "is_enable_bundle",
                                 "bundle"},
219
220
221
222
223
               "eval_at": {"eval_at",
                           "ndcg_eval_at",
                           "ndcg_at",
                           "map_eval_at",
                           "map_at"},
224
225
226
227
228
229
               "group_column": {"group_column",
                                "group",
                                "group_id",
                                "query_column",
                                "query",
                                "query_id"},
230
231
               "header": {"header",
                          "has_header"},
232
233
234
235
236
237
238
239
240
               "ignore_column": {"ignore_column",
                                 "ignore_feature",
                                 "blacklist"},
               "is_enable_sparse": {"is_enable_sparse",
                                    "is_sparse",
                                    "enable_sparse",
                                    "sparse"},
               "label_column": {"label_column",
                                "label"},
241
242
243
               "local_listen_port": {"local_listen_port",
                                     "local_port",
                                     "port"},
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
               "machines": {"machines",
                            "workers",
                            "nodes"},
               "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",
                                  "n_estimators"},
261
262
263
264
265
               "num_threads": {"num_threads",
                               "num_thread",
                               "nthread",
                               "nthreads",
                               "n_jobs"},
266
267
268
269
               "objective": {"objective",
                             "objective_type",
                             "app",
                             "application"},
270
271
               "pre_partition": {"pre_partition",
                                 "is_pre_partition"},
272
273
274
275
               "tree_learner": {"tree_learner",
                                "tree",
                                "tree_type",
                                "tree_learner_type"},
276
277
278
               "two_round": {"two_round",
                             "two_round_loading",
                             "use_two_round_loading"},
279
               "verbosity": {"verbosity",
280
281
282
                             "verbose"},
               "weight_column": {"weight_column",
                                 "weight"}}
283
284
285
286
287

    @classmethod
    def get(cls, *args):
        ret = set()
        for i in args:
288
            ret |= cls.aliases.get(i, {i})
289
290
291
        return ret


292
293
MAX_INT32 = (1 << 31) - 1

294
"""Macro definition of data type in C API of LightGBM"""
wxchan's avatar
wxchan committed
295
296
297
298
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
299

300
"""Matrix is row major in Python"""
wxchan's avatar
wxchan committed
301
302
C_API_IS_ROW_MAJOR = 1

303
"""Macro definition of prediction type in C API of LightGBM"""
wxchan's avatar
wxchan committed
304
305
306
C_API_PREDICT_NORMAL = 0
C_API_PREDICT_RAW_SCORE = 1
C_API_PREDICT_LEAF_INDEX = 2
307
C_API_PREDICT_CONTRIB = 3
wxchan's avatar
wxchan committed
308

309
310
311
312
"""Macro definition of sparse matrix type"""
C_API_MATRIX_TYPE_CSR = 0
C_API_MATRIX_TYPE_CSC = 1

313
314
315
316
"""Macro definition of feature importance type"""
C_API_FEATURE_IMPORTANCE_SPLIT = 0
C_API_FEATURE_IMPORTANCE_GAIN = 1

317
"""Data type of data field"""
wxchan's avatar
wxchan committed
318
319
FIELD_TYPE_MAPPER = {"label": C_API_DTYPE_FLOAT32,
                     "weight": C_API_DTYPE_FLOAT32,
Guolin Ke's avatar
Guolin Ke committed
320
                     "init_score": C_API_DTYPE_FLOAT64,
321
                     "group": C_API_DTYPE_INT32}
wxchan's avatar
wxchan committed
322

323
324
325
326
"""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
327

328
def convert_from_sliced_object(data):
329
    """Fix the memory of multi-dimensional sliced object."""
330
    if isinstance(data, np.ndarray) and isinstance(data.base, np.ndarray):
331
        if not data.flags.c_contiguous:
332
333
            warnings.warn("Usage of np.ndarray subset (sliced data) is not recommended "
                          "due to it will double the peak memory cost in LightGBM.")
334
335
336
337
            return np.copy(data)
    return data


wxchan's avatar
wxchan committed
338
def c_float_array(data):
339
    """Get pointer of float numpy array / list."""
wxchan's avatar
wxchan committed
340
341
342
    if is_1d_list(data):
        data = np.array(data, copy=False)
    if is_numpy_1d_array(data):
343
344
        data = convert_from_sliced_object(data)
        assert data.flags.c_contiguous
wxchan's avatar
wxchan committed
345
346
347
348
349
350
351
        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:
352
            raise TypeError("Expected np.float32 or np.float64, met type({})"
wxchan's avatar
wxchan committed
353
354
                            .format(data.dtype))
    else:
355
        raise TypeError("Unknown type({})".format(type(data).__name__))
356
    return (ptr_data, type_data, data)  # return `data` to avoid the temporary copy is freed
wxchan's avatar
wxchan committed
357

wxchan's avatar
wxchan committed
358

wxchan's avatar
wxchan committed
359
def c_int_array(data):
360
    """Get pointer of int numpy array / list."""
wxchan's avatar
wxchan committed
361
362
363
    if is_1d_list(data):
        data = np.array(data, copy=False)
    if is_numpy_1d_array(data):
364
365
        data = convert_from_sliced_object(data)
        assert data.flags.c_contiguous
wxchan's avatar
wxchan committed
366
367
368
369
370
371
372
        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:
373
            raise TypeError("Expected np.int32 or np.int64, met type({})"
wxchan's avatar
wxchan committed
374
375
                            .format(data.dtype))
    else:
376
        raise TypeError("Unknown type({})".format(type(data).__name__))
377
    return (ptr_data, type_data, data)  # return `data` to avoid the temporary copy is freed
wxchan's avatar
wxchan committed
378

wxchan's avatar
wxchan committed
379

380
381
382
383
384
385
386
387
388
389
390
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


391
def _data_from_pandas(data, feature_name, categorical_feature, pandas_categorical):
392
    if isinstance(data, DataFrame):
393
394
        if len(data.shape) != 2 or data.shape[0] < 1:
            raise ValueError('Input data must be 2 dimensional and non empty.')
395
396
        if feature_name == 'auto' or feature_name is None:
            data = data.rename(columns=str)
397
398
        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]
399
400
401
402
403
        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.')
404
            for col, category in zip(cat_cols, pandas_categorical):
405
406
                if list(data[col].cat.categories) != list(category):
                    data[col] = data[col].cat.set_categories(category)
407
        if len(cat_cols):  # cat_cols is list
408
            data = data.copy()  # not alter origin DataFrame
409
            data[cat_cols] = data[cat_cols].apply(lambda x: x.cat.codes).replace({-1: np.nan})
410
411
412
        if categorical_feature is not None:
            if feature_name is None:
                feature_name = list(data.columns)
413
            if categorical_feature == 'auto':  # use cat cols from DataFrame
414
                categorical_feature = cat_cols_not_ordered
415
416
            else:  # use cat cols specified by user
                categorical_feature = list(categorical_feature)
417
418
        if feature_name == 'auto':
            feature_name = list(data.columns)
419
420
        bad_indices = _get_bad_pandas_dtypes(data.dtypes)
        if bad_indices:
421
            raise ValueError("DataFrame.dtypes for data must be int, float or bool.\n"
422
                             "Did not expect the data types in the following fields: "
423
                             + ', '.join(data.columns[bad_indices]))
424
425
426
        data = data.values
        if data.dtype != np.float32 and data.dtype != np.float64:
            data = data.astype(np.float32)
427
428
429
430
431
432
    else:
        if feature_name == 'auto':
            feature_name = None
        if categorical_feature == 'auto':
            categorical_feature = None
    return data, feature_name, categorical_feature, pandas_categorical
433
434
435
436
437
438


def _label_from_pandas(label):
    if isinstance(label, DataFrame):
        if len(label.columns) > 1:
            raise ValueError('DataFrame for label cannot have multiple columns')
439
        if _get_bad_pandas_dtypes(label.dtypes):
440
            raise ValueError('DataFrame.dtypes for label must be int, float or bool')
441
        label = np.ravel(label.values.astype(np.float32, copy=False))
442
443
444
    return label


445
446
447
448
449
450
451
452
453
454
455
def _dump_pandas_categorical(pandas_categorical, file_name=None):
    pandas_str = ('\npandas_categorical:'
                  + json.dumps(pandas_categorical, default=json_default_with_numpy)
                  + '\n')
    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):
456
457
    pandas_key = 'pandas_categorical:'
    offset = -len(pandas_key)
458
    if file_name is not None:
459
460
461
462
463
464
465
466
467
468
        max_offset = -os.path.getsize(file_name)
        with open(file_name, 'rb') as f:
            while True:
                if offset < max_offset:
                    offset = max_offset
                f.seek(offset, os.SEEK_END)
                lines = f.readlines()
                if len(lines) >= 2:
                    break
                offset *= 2
469
        last_line = lines[-1].decode('utf-8').strip()
470
        if not last_line.startswith(pandas_key):
471
            last_line = lines[-2].decode('utf-8').strip()
472
    elif model_str is not None:
473
474
475
476
477
478
        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
479
480


481
class _InnerPredictor:
482
483
484
485
486
    """_InnerPredictor of LightGBM.

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

Nikita Titov's avatar
Nikita Titov committed
487
488
489
    .. note::

        Can be converted from Booster, but cannot be converted to Booster.
Guolin Ke's avatar
Guolin Ke committed
490
    """
491

492
    def __init__(self, model_file=None, booster_handle=None, pred_parameter=None):
493
        """Initialize the _InnerPredictor.
wxchan's avatar
wxchan committed
494
495
496

        Parameters
        ----------
497
        model_file : string or None, optional (default=None)
wxchan's avatar
wxchan committed
498
            Path to the model file.
499
500
501
502
        booster_handle : object or None, optional (default=None)
            Handle of Booster.
        pred_parameter: dict or None, optional (default=None)
            Other parameters for the prediciton.
wxchan's avatar
wxchan committed
503
504
505
506
507
        """
        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
508
            out_num_iterations = ctypes.c_int(0)
wxchan's avatar
wxchan committed
509
510
511
512
            _safe_call(_LIB.LGBM_BoosterCreateFromModelfile(
                c_str(model_file),
                ctypes.byref(out_num_iterations),
                ctypes.byref(self.handle)))
Guolin Ke's avatar
Guolin Ke committed
513
            out_num_class = ctypes.c_int(0)
wxchan's avatar
wxchan committed
514
515
516
517
            _safe_call(_LIB.LGBM_BoosterGetNumClasses(
                self.handle,
                ctypes.byref(out_num_class)))
            self.num_class = out_num_class.value
518
            self.num_total_iteration = out_num_iterations.value
519
            self.pandas_categorical = _load_pandas_categorical(file_name=model_file)
wxchan's avatar
wxchan committed
520
        elif booster_handle is not None:
Guolin Ke's avatar
Guolin Ke committed
521
            self.__is_manage_handle = False
wxchan's avatar
wxchan committed
522
            self.handle = booster_handle
Guolin Ke's avatar
Guolin Ke committed
523
            out_num_class = ctypes.c_int(0)
wxchan's avatar
wxchan committed
524
525
526
527
            _safe_call(_LIB.LGBM_BoosterGetNumClasses(
                self.handle,
                ctypes.byref(out_num_class)))
            self.num_class = out_num_class.value
528
            self.num_total_iteration = self.current_iteration()
529
            self.pandas_categorical = None
wxchan's avatar
wxchan committed
530
        else:
531
            raise TypeError('Need model_file or booster_handle to create a predictor')
wxchan's avatar
wxchan committed
532

533
534
        pred_parameter = {} if pred_parameter is None else pred_parameter
        self.pred_parameter = param_dict_to_str(pred_parameter)
cbecker's avatar
cbecker committed
535

wxchan's avatar
wxchan committed
536
    def __del__(self):
537
538
539
540
541
        try:
            if self.__is_manage_handle:
                _safe_call(_LIB.LGBM_BoosterFree(self.handle))
        except AttributeError:
            pass
wxchan's avatar
wxchan committed
542

543
544
545
546
547
    def __getstate__(self):
        this = self.__dict__.copy()
        this.pop('handle', None)
        return this

548
    def predict(self, data, start_iteration=0, num_iteration=-1,
549
                raw_score=False, pred_leaf=False, pred_contrib=False, data_has_header=False,
wxchan's avatar
wxchan committed
550
                is_reshape=True):
551
        """Predict logic.
wxchan's avatar
wxchan committed
552
553
554

        Parameters
        ----------
555
        data : string, numpy array, pandas DataFrame, H2O DataTable's Frame or scipy.sparse
556
557
            Data source for prediction.
            When data type is string, it represents the path of txt file.
558
559
        start_iteration : int, optional (default=0)
            Start index of the iteration to predict.
560
561
562
563
564
565
566
567
568
569
570
571
572
        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
573
574
575

        Returns
        -------
576
        result : numpy array, scipy.sparse or list of scipy.sparse
577
            Prediction result.
578
            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
579
        """
wxchan's avatar
wxchan committed
580
        if isinstance(data, Dataset):
581
            raise TypeError("Cannot use Dataset instance for prediction, please use raw data instead")
582
        data = _data_from_pandas(data, None, None, self.pandas_categorical)[0]
wxchan's avatar
wxchan committed
583
584
585
586
587
        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
588
589
        if pred_contrib:
            predict_type = C_API_PREDICT_CONTRIB
wxchan's avatar
wxchan committed
590
        int_data_has_header = 1 if data_has_header else 0
cbecker's avatar
cbecker committed
591

592
        if isinstance(data, str):
593
            with _TempFile() as f:
wxchan's avatar
wxchan committed
594
595
596
                _safe_call(_LIB.LGBM_BoosterPredictForFile(
                    self.handle,
                    c_str(data),
Guolin Ke's avatar
Guolin Ke committed
597
598
                    ctypes.c_int(int_data_has_header),
                    ctypes.c_int(predict_type),
599
                    ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
600
                    ctypes.c_int(num_iteration),
601
                    c_str(self.pred_parameter),
wxchan's avatar
wxchan committed
602
603
                    c_str(f.name)))
                lines = f.readlines()
604
605
                nrow = len(lines)
                preds = [float(token) for line in lines for token in line.split('\t')]
Guolin Ke's avatar
Guolin Ke committed
606
                preds = np.array(preds, dtype=np.float64, copy=False)
wxchan's avatar
wxchan committed
607
        elif isinstance(data, scipy.sparse.csr_matrix):
608
            preds, nrow = self.__pred_for_csr(data, start_iteration, num_iteration, predict_type)
Guolin Ke's avatar
Guolin Ke committed
609
        elif isinstance(data, scipy.sparse.csc_matrix):
610
            preds, nrow = self.__pred_for_csc(data, start_iteration, num_iteration, predict_type)
wxchan's avatar
wxchan committed
611
        elif isinstance(data, np.ndarray):
612
            preds, nrow = self.__pred_for_np2d(data, start_iteration, num_iteration, predict_type)
613
614
615
        elif isinstance(data, list):
            try:
                data = np.array(data)
616
            except BaseException:
617
                raise ValueError('Cannot convert data list to numpy array.')
618
            preds, nrow = self.__pred_for_np2d(data, start_iteration, num_iteration, predict_type)
619
        elif isinstance(data, DataTable):
620
            preds, nrow = self.__pred_for_np2d(data.to_numpy(), start_iteration, num_iteration, predict_type)
wxchan's avatar
wxchan committed
621
622
        else:
            try:
623
                warnings.warn('Converting data to scipy sparse matrix.')
wxchan's avatar
wxchan committed
624
                csr = scipy.sparse.csr_matrix(data)
625
            except BaseException:
626
                raise TypeError('Cannot predict data for type {}'.format(type(data).__name__))
627
            preds, nrow = self.__pred_for_csr(csr, start_iteration, num_iteration, predict_type)
wxchan's avatar
wxchan committed
628
629
        if pred_leaf:
            preds = preds.astype(np.int32)
630
631
        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
632
            if preds.size % nrow == 0:
633
                preds = preds.reshape(nrow, -1)
wxchan's avatar
wxchan committed
634
            else:
635
                raise ValueError('Length of predict result (%d) cannot be divide nrow (%d)'
wxchan's avatar
wxchan committed
636
637
638
                                 % (preds.size, nrow))
        return preds

639
    def __get_num_preds(self, start_iteration, num_iteration, nrow, predict_type):
640
        """Get size of prediction result."""
641
642
643
644
645
        if nrow > MAX_INT32:
            raise LightGBMError('LightGBM cannot perform prediction for data'
                                'with number of rows greater than MAX_INT32 (%d).\n'
                                'You can split your data into chunks'
                                'and then concatenate predictions for them' % MAX_INT32)
Guolin Ke's avatar
Guolin Ke committed
646
647
648
        n_preds = ctypes.c_int64(0)
        _safe_call(_LIB.LGBM_BoosterCalcNumPredict(
            self.handle,
Guolin Ke's avatar
Guolin Ke committed
649
650
            ctypes.c_int(nrow),
            ctypes.c_int(predict_type),
651
            ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
652
            ctypes.c_int(num_iteration),
Guolin Ke's avatar
Guolin Ke committed
653
654
            ctypes.byref(n_preds)))
        return n_preds.value
wxchan's avatar
wxchan committed
655

656
    def __pred_for_np2d(self, mat, start_iteration, num_iteration, predict_type):
657
        """Predict for a 2-D numpy matrix."""
wxchan's avatar
wxchan committed
658
        if len(mat.shape) != 2:
659
            raise ValueError('Input numpy.ndarray or list must be 2 dimensional')
wxchan's avatar
wxchan committed
660

661
        def inner_predict(mat, start_iteration, num_iteration, predict_type, preds=None):
662
663
            if mat.dtype == np.float32 or mat.dtype == np.float64:
                data = np.array(mat.reshape(mat.size), dtype=mat.dtype, copy=False)
664
            else:  # change non-float data to float data, need to copy
665
666
                data = np.array(mat.reshape(mat.size), dtype=np.float32)
            ptr_data, type_ptr_data, _ = c_float_array(data)
667
            n_preds = self.__get_num_preds(start_iteration, num_iteration, mat.shape[0], predict_type)
668
669
670
671
672
673
674
675
676
677
678
679
680
            if preds is None:
                preds = np.zeros(n_preds, dtype=np.float64)
            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),
                ctypes.c_int(mat.shape[0]),
                ctypes.c_int(mat.shape[1]),
                ctypes.c_int(C_API_IS_ROW_MAJOR),
                ctypes.c_int(predict_type),
681
                ctypes.c_int(start_iteration),
682
683
684
685
686
687
688
689
690
691
692
693
                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
694
            n_preds = [self.__get_num_preds(start_iteration, num_iteration, i, predict_type) for i in np.diff([0] + list(sections) + [nrow])]
695
696
            n_preds_sections = np.array([0] + n_preds, dtype=np.intp).cumsum()
            preds = np.zeros(sum(n_preds), dtype=np.float64)
697
698
            for chunk, (start_idx_pred, end_idx_pred) in zip(np.array_split(mat, sections),
                                                             zip(n_preds_sections, n_preds_sections[1:])):
699
                # avoid memory consumption by arrays concatenation operations
700
                inner_predict(chunk, start_iteration, num_iteration, predict_type, preds[start_idx_pred:end_idx_pred])
701
            return preds, nrow
wxchan's avatar
wxchan committed
702
        else:
703
            return inner_predict(mat, start_iteration, num_iteration, predict_type)
wxchan's avatar
wxchan committed
704

705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
    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

751
    def __pred_for_csr(self, csr, start_iteration, num_iteration, predict_type):
752
        """Predict for a CSR data."""
753
        def inner_predict(csr, start_iteration, num_iteration, predict_type, preds=None):
754
            nrow = len(csr.indptr) - 1
755
            n_preds = self.__get_num_preds(start_iteration, num_iteration, nrow, predict_type)
756
757
758
759
760
761
762
763
764
            if preds is None:
                preds = np.zeros(n_preds, dtype=np.float64)
            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)

765
            assert csr.shape[1] <= MAX_INT32
766
            csr_indices = csr.indices.astype(np.int32, copy=False)
767

768
769
770
771
            _safe_call(_LIB.LGBM_BoosterPredictForCSR(
                self.handle,
                ptr_indptr,
                ctypes.c_int32(type_ptr_indptr),
772
                csr_indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
773
774
775
776
777
778
                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),
779
                ctypes.c_int(start_iteration),
780
781
782
783
784
785
786
                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
787

788
        def inner_predict_sparse(csr, start_iteration, num_iteration, predict_type):
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
            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)()
            out_shape = np.zeros(2, dtype=np.int64)
            _safe_call(_LIB.LGBM_BoosterPredictSparseOutput(
                self.handle,
                ptr_indptr,
                ctypes.c_int32(type_ptr_indptr),
                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),
814
                ctypes.c_int(start_iteration),
815
816
817
818
819
820
821
822
823
824
825
826
827
                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:
828
            return inner_predict_sparse(csr, start_iteration, num_iteration, predict_type)
829
830
831
832
        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
833
            n_preds = [self.__get_num_preds(start_iteration, num_iteration, i, predict_type) for i in np.diff(sections)]
834
835
            n_preds_sections = np.array([0] + n_preds, dtype=np.intp).cumsum()
            preds = np.zeros(sum(n_preds), dtype=np.float64)
836
837
            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:])):
838
                # avoid memory consumption by arrays concatenation operations
839
                inner_predict(csr[start_idx:end_idx], start_iteration, num_iteration, predict_type, preds[start_idx_pred:end_idx_pred])
840
841
            return preds, nrow
        else:
842
            return inner_predict(csr, start_iteration, num_iteration, predict_type)
Guolin Ke's avatar
Guolin Ke committed
843

844
    def __pred_for_csc(self, csc, start_iteration, num_iteration, predict_type):
845
        """Predict for a CSC data."""
846
        def inner_predict_sparse(csc, start_iteration, num_iteration, predict_type):
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
            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)()
            out_shape = np.zeros(2, dtype=np.int64)
            _safe_call(_LIB.LGBM_BoosterPredictSparseOutput(
                self.handle,
                ptr_indptr,
                ctypes.c_int32(type_ptr_indptr),
                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),
872
                ctypes.c_int(start_iteration),
873
874
875
876
877
878
879
880
881
882
883
884
                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
885
        nrow = csc.shape[0]
886
        if nrow > MAX_INT32:
887
            return self.__pred_for_csr(csc.tocsr(), start_iteration, num_iteration, predict_type)
888
        if predict_type == C_API_PREDICT_CONTRIB:
889
890
            return inner_predict_sparse(csc, start_iteration, num_iteration, predict_type)
        n_preds = self.__get_num_preds(start_iteration, num_iteration, nrow, predict_type)
Guolin Ke's avatar
Guolin Ke committed
891
892
893
        preds = np.zeros(n_preds, dtype=np.float64)
        out_num_preds = ctypes.c_int64(0)

894
895
        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
896

897
        assert csc.shape[0] <= MAX_INT32
898
        csc_indices = csc.indices.astype(np.int32, copy=False)
899

Guolin Ke's avatar
Guolin Ke committed
900
901
902
        _safe_call(_LIB.LGBM_BoosterPredictForCSC(
            self.handle,
            ptr_indptr,
Guolin Ke's avatar
Guolin Ke committed
903
            ctypes.c_int32(type_ptr_indptr),
904
            csc_indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
Guolin Ke's avatar
Guolin Ke committed
905
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
906
907
908
909
910
            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),
911
            ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
912
            ctypes.c_int(num_iteration),
913
            c_str(self.pred_parameter),
Guolin Ke's avatar
Guolin Ke committed
914
            ctypes.byref(out_num_preds),
wxchan's avatar
wxchan committed
915
            preds.ctypes.data_as(ctypes.POINTER(ctypes.c_double))))
wxchan's avatar
wxchan committed
916
        if n_preds != out_num_preds.value:
917
            raise ValueError("Wrong length for predict results")
wxchan's avatar
wxchan committed
918
919
        return preds, nrow

920
921
922
923
924
925
926
927
928
929
930
931
932
933
    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
934

935
class Dataset:
wxchan's avatar
wxchan committed
936
    """Dataset in LightGBM."""
937

938
    def __init__(self, data, label=None, reference=None,
939
                 weight=None, group=None, init_score=None, silent=False,
940
                 feature_name='auto', categorical_feature='auto', params=None,
wxchan's avatar
wxchan committed
941
                 free_raw_data=True):
942
        """Initialize Dataset.
943

wxchan's avatar
wxchan committed
944
945
        Parameters
        ----------
946
        data : string, numpy array, pandas DataFrame, H2O DataTable's Frame, scipy.sparse or list of numpy arrays
wxchan's avatar
wxchan committed
947
            Data source of Dataset.
948
            If string, it represents the path to txt file.
949
        label : list, numpy 1-D array, pandas Series / one-column DataFrame or None, optional (default=None)
950
951
952
            Label of the data.
        reference : Dataset or None, optional (default=None)
            If this is Dataset for validation, training data should be used as reference.
953
        weight : list, numpy 1-D array, pandas Series or None, optional (default=None)
wxchan's avatar
wxchan committed
954
            Weight for each instance.
955
        group : list, numpy 1-D array, pandas Series or None, optional (default=None)
956
957
958
            Group/query data.
            Only used in the learning-to-rank task.
            sum(group) = n_samples.
959
960
            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.
961
        init_score : list, numpy 1-D array, pandas Series or None, optional (default=None)
962
            Init score for Dataset.
963
964
965
966
967
968
969
970
971
        silent : bool, optional (default=False)
            Whether to print messages during construction.
        feature_name : list of strings or 'auto', optional (default="auto")
            Feature names.
            If 'auto' and data is pandas DataFrame, data columns names are used.
        categorical_feature : list of strings or int, or 'auto', optional (default="auto")
            Categorical features.
            If list of int, interpreted as indices.
            If list of strings, interpreted as feature names (need to specify ``feature_name`` as well).
972
            If 'auto' and data is pandas DataFrame, pandas unordered categorical columns are used.
973
            All values in categorical features should be less than int32 max value (2147483647).
974
            Large values could be memory consuming. Consider using consecutive integers starting from zero.
975
            All negative values in categorical features will be treated as missing values.
976
            The output cannot be monotonically constrained with respect to a categorical feature.
Nikita Titov's avatar
Nikita Titov committed
977
        params : dict or None, optional (default=None)
978
            Other parameters for Dataset.
Nikita Titov's avatar
Nikita Titov committed
979
        free_raw_data : bool, optional (default=True)
980
            If True, raw data is freed after constructing inner Dataset.
wxchan's avatar
wxchan committed
981
        """
wxchan's avatar
wxchan committed
982
983
984
985
986
987
        self.handle = None
        self.data = data
        self.label = label
        self.reference = reference
        self.weight = weight
        self.group = group
988
        self.init_score = init_score
wxchan's avatar
wxchan committed
989
990
        self.silent = silent
        self.feature_name = feature_name
991
        self.categorical_feature = categorical_feature
992
        self.params = copy.deepcopy(params)
wxchan's avatar
wxchan committed
993
994
        self.free_raw_data = free_raw_data
        self.used_indices = None
995
        self.need_slice = True
wxchan's avatar
wxchan committed
996
        self._predictor = None
997
        self.pandas_categorical = None
998
        self.params_back_up = None
999
1000
        self.feature_penalty = None
        self.monotone_constraints = None
1001
        self.version = 0
wxchan's avatar
wxchan committed
1002
1003

    def __del__(self):
1004
1005
1006
1007
        try:
            self._free_handle()
        except AttributeError:
            pass
1008

1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
    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",
1030
                                                "linear_tree",
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
                                                "max_bin",
                                                "max_bin_by_feature",
                                                "min_data_in_bin",
                                                "pre_partition",
                                                "two_round",
                                                "use_missing",
                                                "weight_column",
                                                "zero_as_missing")
            return {k: v for k, v in self.params.items() if k in dataset_params}

1041
    def _free_handle(self):
1042
        if self.handle is not None:
1043
            _safe_call(_LIB.LGBM_DatasetFree(self.handle))
1044
            self.handle = None
Guolin Ke's avatar
Guolin Ke committed
1045
1046
1047
        self.need_slice = True
        if self.used_indices is not None:
            self.data = None
Nikita Titov's avatar
Nikita Titov committed
1048
        return self
wxchan's avatar
wxchan committed
1049

Guolin Ke's avatar
Guolin Ke committed
1050
1051
    def _set_init_score_by_predictor(self, predictor, data, used_indices=None):
        data_has_header = False
1052
        if isinstance(data, str):
Guolin Ke's avatar
Guolin Ke committed
1053
            # check data has header or not
1054
            data_has_header = any(self.params.get(alias, False) for alias in _ConfigAliases.get("header"))
Guolin Ke's avatar
Guolin Ke committed
1055
        num_data = self.num_data()
1056
1057
1058
1059
1060
1061
1062
        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
1063
                if isinstance(data, str):
1064
1065
                    sub_init_score = np.zeros(num_data * predictor.num_class, dtype=np.float32)
                    assert num_data == len(used_indices)
1066
1067
                    for i in range(len(used_indices)):
                        for j in range(predictor.num_class):
1068
1069
1070
1071
1072
                            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
                new_init_score = np.zeros(init_score.size, dtype=np.float32)
1073
1074
                for i in range(num_data):
                    for j in range(predictor.num_class):
1075
1076
1077
1078
1079
1080
                        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:
            init_score = np.zeros(self.init_score.shape, dtype=np.float32)
        else:
            return self
Guolin Ke's avatar
Guolin Ke committed
1081
1082
        self.set_init_score(init_score)

1083
    def _lazy_init(self, data, label=None, reference=None,
1084
                   weight=None, group=None, init_score=None, predictor=None,
wxchan's avatar
wxchan committed
1085
                   silent=False, feature_name='auto',
1086
                   categorical_feature='auto', params=None):
wxchan's avatar
wxchan committed
1087
1088
        if data is None:
            self.handle = None
Nikita Titov's avatar
Nikita Titov committed
1089
            return self
Guolin Ke's avatar
Guolin Ke committed
1090
1091
1092
        if reference is not None:
            self.pandas_categorical = reference.pandas_categorical
            categorical_feature = reference.categorical_feature
1093
1094
1095
1096
        data, feature_name, categorical_feature, self.pandas_categorical = _data_from_pandas(data,
                                                                                             feature_name,
                                                                                             categorical_feature,
                                                                                             self.pandas_categorical)
wxchan's avatar
wxchan committed
1097
        label = _label_from_pandas(label)
Guolin Ke's avatar
Guolin Ke committed
1098

1099
        # process for args
wxchan's avatar
wxchan committed
1100
        params = {} if params is None else params
1101
1102
1103
        args_names = (getattr(self.__class__, '_lazy_init')
                      .__code__
                      .co_varnames[:getattr(self.__class__, '_lazy_init').__code__.co_argcount])
1104
1105
        for key, _ in params.items():
            if key in args_names:
1106
1107
1108
                warnings.warn('{0} keyword has been found in `params` and will be ignored.\n'
                              'Please use {0} argument of the Dataset constructor to pass this parameter.'
                              .format(key))
1109
        # user can set verbose with params, it has higher priority
1110
        if not any(verbose_alias in params for verbose_alias in _ConfigAliases.get("verbosity")) and silent:
1111
            params["verbose"] = -1
1112
        # get categorical features
1113
1114
1115
1116
1117
1118
        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:
1119
                if isinstance(name, str) and name in feature_dict:
1120
                    categorical_indices.add(feature_dict[name])
1121
                elif isinstance(name, int):
1122
1123
1124
1125
                    categorical_indices.add(name)
                else:
                    raise TypeError("Wrong type({}) or unknown name({}) in categorical_feature"
                                    .format(type(name).__name__, name))
1126
            if categorical_indices:
1127
1128
1129
1130
                for cat_alias in _ConfigAliases.get("categorical_feature"):
                    if cat_alias in params:
                        warnings.warn('{} in param dict is overridden.'.format(cat_alias))
                        params.pop(cat_alias, None)
1131
                params['categorical_column'] = sorted(categorical_indices)
1132

wxchan's avatar
wxchan committed
1133
        params_str = param_dict_to_str(params)
1134
        self.params = params
1135
        # process for reference dataset
wxchan's avatar
wxchan committed
1136
        ref_dataset = None
wxchan's avatar
wxchan committed
1137
        if isinstance(reference, Dataset):
1138
            ref_dataset = reference.construct().handle
wxchan's avatar
wxchan committed
1139
1140
        elif reference is not None:
            raise TypeError('Reference dataset should be None or dataset instance')
1141
        # start construct data
1142
        if isinstance(data, str):
wxchan's avatar
wxchan committed
1143
1144
1145
1146
1147
1148
1149
1150
            self.handle = ctypes.c_void_p()
            _safe_call(_LIB.LGBM_DatasetCreateFromFile(
                c_str(data),
                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
1151
1152
        elif isinstance(data, scipy.sparse.csc_matrix):
            self.__init_from_csc(data, params_str, ref_dataset)
wxchan's avatar
wxchan committed
1153
1154
        elif isinstance(data, np.ndarray):
            self.__init_from_np2d(data, params_str, ref_dataset)
1155
1156
        elif isinstance(data, list) and len(data) > 0 and all(isinstance(x, np.ndarray) for x in data):
            self.__init_from_list_np2d(data, params_str, ref_dataset)
1157
1158
        elif isinstance(data, DataTable):
            self.__init_from_np2d(data.to_numpy(), params_str, ref_dataset)
wxchan's avatar
wxchan committed
1159
1160
1161
1162
        else:
            try:
                csr = scipy.sparse.csr_matrix(data)
                self.__init_from_csr(csr, params_str, ref_dataset)
1163
            except BaseException:
wxchan's avatar
wxchan committed
1164
                raise TypeError('Cannot initialize Dataset from {}'.format(type(data).__name__))
wxchan's avatar
wxchan committed
1165
1166
1167
        if label is not None:
            self.set_label(label)
        if self.get_label() is None:
1168
            raise ValueError("Label should not be None")
wxchan's avatar
wxchan committed
1169
1170
1171
1172
        if weight is not None:
            self.set_weight(weight)
        if group is not None:
            self.set_group(group)
1173
1174
1175
        if isinstance(predictor, _InnerPredictor):
            if self._predictor is None and init_score is not None:
                warnings.warn("The init_score will be overridden by the prediction of init_model.")
Guolin Ke's avatar
Guolin Ke committed
1176
            self._set_init_score_by_predictor(predictor, data)
1177
1178
        elif init_score is not None:
            self.set_init_score(init_score)
Guolin Ke's avatar
Guolin Ke committed
1179
1180
        elif predictor is not None:
            raise TypeError('Wrong predictor type {}'.format(type(predictor).__name__))
Guolin Ke's avatar
Guolin Ke committed
1181
        # set feature names
Nikita Titov's avatar
Nikita Titov committed
1182
        return self.set_feature_name(feature_name)
wxchan's avatar
wxchan committed
1183
1184

    def __init_from_np2d(self, mat, params_str, ref_dataset):
1185
        """Initialize data from a 2-D numpy matrix."""
wxchan's avatar
wxchan committed
1186
1187
1188
1189
1190
1191
        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)
1192
        else:  # change non-float data to float data, need to copy
wxchan's avatar
wxchan committed
1193
1194
            data = np.array(mat.reshape(mat.size), dtype=np.float32)

1195
        ptr_data, type_ptr_data, _ = c_float_array(data)
wxchan's avatar
wxchan committed
1196
1197
        _safe_call(_LIB.LGBM_DatasetCreateFromMat(
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
1198
1199
1200
1201
            ctypes.c_int(type_ptr_data),
            ctypes.c_int(mat.shape[0]),
            ctypes.c_int(mat.shape[1]),
            ctypes.c_int(C_API_IS_ROW_MAJOR),
wxchan's avatar
wxchan committed
1202
1203
1204
            c_str(params_str),
            ref_dataset,
            ctypes.byref(self.handle)))
Nikita Titov's avatar
Nikita Titov committed
1205
        return self
wxchan's avatar
wxchan committed
1206

1207
    def __init_from_list_np2d(self, mats, params_str, ref_dataset):
1208
        """Initialize data from a list of 2-D numpy matrices."""
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
        ncol = mats[0].shape[1]
        nrow = np.zeros((len(mats),), np.int32)
        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)
1230
            else:  # change non-float data to float data, need to copy
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
                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(
            ctypes.c_int(len(mats)),
            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)),
            ctypes.c_int(ncol),
            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
1251
        return self
1252

wxchan's avatar
wxchan committed
1253
    def __init_from_csr(self, csr, params_str, ref_dataset):
1254
        """Initialize data from a CSR matrix."""
wxchan's avatar
wxchan committed
1255
        if len(csr.indices) != len(csr.data):
1256
            raise ValueError('Length mismatch: {} vs {}'.format(len(csr.indices), len(csr.data)))
wxchan's avatar
wxchan committed
1257
1258
        self.handle = ctypes.c_void_p()

1259
1260
        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
1261

1262
        assert csr.shape[1] <= MAX_INT32
1263
        csr_indices = csr.indices.astype(np.int32, copy=False)
1264

wxchan's avatar
wxchan committed
1265
1266
        _safe_call(_LIB.LGBM_DatasetCreateFromCSR(
            ptr_indptr,
Guolin Ke's avatar
Guolin Ke committed
1267
            ctypes.c_int(type_ptr_indptr),
1268
            csr_indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
wxchan's avatar
wxchan committed
1269
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
1270
1271
1272
1273
            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
1274
1275
1276
            c_str(params_str),
            ref_dataset,
            ctypes.byref(self.handle)))
Nikita Titov's avatar
Nikita Titov committed
1277
        return self
wxchan's avatar
wxchan committed
1278

Guolin Ke's avatar
Guolin Ke committed
1279
    def __init_from_csc(self, csc, params_str, ref_dataset):
1280
        """Initialize data from a CSC matrix."""
Guolin Ke's avatar
Guolin Ke committed
1281
1282
1283
1284
        if len(csc.indices) != len(csc.data):
            raise ValueError('Length mismatch: {} vs {}'.format(len(csc.indices), len(csc.data)))
        self.handle = ctypes.c_void_p()

1285
1286
        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
1287

1288
        assert csc.shape[0] <= MAX_INT32
1289
        csc_indices = csc.indices.astype(np.int32, copy=False)
1290

Guolin Ke's avatar
Guolin Ke committed
1291
1292
        _safe_call(_LIB.LGBM_DatasetCreateFromCSC(
            ptr_indptr,
Guolin Ke's avatar
Guolin Ke committed
1293
            ctypes.c_int(type_ptr_indptr),
1294
            csc_indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
Guolin Ke's avatar
Guolin Ke committed
1295
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
1296
1297
1298
1299
            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
1300
1301
1302
            c_str(params_str),
            ref_dataset,
            ctypes.byref(self.handle)))
Nikita Titov's avatar
Nikita Titov committed
1303
        return self
Guolin Ke's avatar
Guolin Ke committed
1304

wxchan's avatar
wxchan committed
1305
    def construct(self):
1306
1307
1308
1309
1310
        """Lazy init.

        Returns
        -------
        self : Dataset
Nikita Titov's avatar
Nikita Titov committed
1311
            Constructed Dataset object.
1312
        """
1313
        if self.handle is None:
wxchan's avatar
wxchan committed
1314
            if self.reference is not None:
1315
1316
1317
1318
                reference_params = self.reference.get_params()
                if self.get_params() != reference_params:
                    warnings.warn('Overriding the parameters from Reference Dataset.')
                    self._update_params(reference_params)
wxchan's avatar
wxchan committed
1319
                if self.used_indices is None:
1320
                    # create valid
1321
                    self._lazy_init(self.data, label=self.label, reference=self.reference,
1322
1323
                                    weight=self.weight, group=self.group,
                                    init_score=self.init_score, predictor=self._predictor,
1324
                                    silent=self.silent, feature_name=self.feature_name, params=self.params)
wxchan's avatar
wxchan committed
1325
                else:
1326
                    # construct subset
wxchan's avatar
wxchan committed
1327
                    used_indices = list_to_1d_numpy(self.used_indices, np.int32, name='used_indices')
1328
                    assert used_indices.flags.c_contiguous
Guolin Ke's avatar
Guolin Ke committed
1329
                    if self.reference.group is not None:
1330
                        group_info = np.array(self.reference.group).astype(np.int32, copy=False)
1331
                        _, self.group = np.unique(np.repeat(range(len(group_info)), repeats=group_info)[self.used_indices],
1332
                                                  return_counts=True)
1333
                    self.handle = ctypes.c_void_p()
wxchan's avatar
wxchan committed
1334
1335
                    params_str = param_dict_to_str(self.params)
                    _safe_call(_LIB.LGBM_DatasetGetSubset(
1336
                        self.reference.construct().handle,
wxchan's avatar
wxchan committed
1337
                        used_indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
Guolin Ke's avatar
Guolin Ke committed
1338
                        ctypes.c_int(used_indices.shape[0]),
wxchan's avatar
wxchan committed
1339
1340
                        c_str(params_str),
                        ctypes.byref(self.handle)))
Guolin Ke's avatar
Guolin Ke committed
1341
1342
                    if not self.free_raw_data:
                        self.get_data()
Guolin Ke's avatar
Guolin Ke committed
1343
1344
                    if self.group is not None:
                        self.set_group(self.group)
wxchan's avatar
wxchan committed
1345
1346
                    if self.get_label() is None:
                        raise ValueError("Label should not be None.")
Guolin Ke's avatar
Guolin Ke committed
1347
1348
1349
                    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
1350
            else:
1351
                # create train
1352
                self._lazy_init(self.data, label=self.label,
1353
1354
1355
                                weight=self.weight, group=self.group,
                                init_score=self.init_score, predictor=self._predictor,
                                silent=self.silent, feature_name=self.feature_name,
1356
                                categorical_feature=self.categorical_feature, params=self.params)
wxchan's avatar
wxchan committed
1357
1358
1359
            if self.free_raw_data:
                self.data = None
        return self
wxchan's avatar
wxchan committed
1360

wxchan's avatar
wxchan committed
1361
    def create_valid(self, data, label=None, weight=None, group=None,
1362
                     init_score=None, silent=False, params=None):
1363
        """Create validation data align with current Dataset.
wxchan's avatar
wxchan committed
1364
1365
1366

        Parameters
        ----------
1367
        data : string, numpy array, pandas DataFrame, H2O DataTable's Frame, scipy.sparse or list of numpy arrays
wxchan's avatar
wxchan committed
1368
            Data source of Dataset.
1369
            If string, it represents the path to txt file.
1370
        label : list, numpy 1-D array, pandas Series / one-column DataFrame or None, optional (default=None)
1371
1372
            Label of the data.
        weight : list, numpy 1-D array, pandas Series or None, optional (default=None)
wxchan's avatar
wxchan committed
1373
            Weight for each instance.
1374
        group : list, numpy 1-D array, pandas Series or None, optional (default=None)
1375
1376
1377
            Group/query data.
            Only used in the learning-to-rank task.
            sum(group) = n_samples.
1378
1379
            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.
1380
        init_score : list, numpy 1-D array, pandas Series or None, optional (default=None)
1381
            Init score for Dataset.
1382
1383
        silent : bool, optional (default=False)
            Whether to print messages during construction.
Nikita Titov's avatar
Nikita Titov committed
1384
        params : dict or None, optional (default=None)
1385
            Other parameters for validation Dataset.
1386
1387
1388

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1389
1390
        valid : Dataset
            Validation Dataset with reference to self.
wxchan's avatar
wxchan committed
1391
        """
1392
        ret = Dataset(data, label=label, reference=self,
1393
1394
                      weight=weight, group=group, init_score=init_score,
                      silent=silent, params=params, free_raw_data=self.free_raw_data)
wxchan's avatar
wxchan committed
1395
        ret._predictor = self._predictor
1396
        ret.pandas_categorical = self.pandas_categorical
wxchan's avatar
wxchan committed
1397
        return ret
wxchan's avatar
wxchan committed
1398

wxchan's avatar
wxchan committed
1399
    def subset(self, used_indices, params=None):
1400
        """Get subset of current Dataset.
wxchan's avatar
wxchan committed
1401
1402
1403
1404

        Parameters
        ----------
        used_indices : list of int
1405
            Indices used to create the subset.
Nikita Titov's avatar
Nikita Titov committed
1406
        params : dict or None, optional (default=None)
1407
            These parameters will be passed to Dataset constructor.
1408
1409
1410
1411
1412

        Returns
        -------
        subset : Dataset
            Subset of the current Dataset.
wxchan's avatar
wxchan committed
1413
        """
wxchan's avatar
wxchan committed
1414
1415
        if params is None:
            params = self.params
wxchan's avatar
wxchan committed
1416
        ret = Dataset(None, reference=self, feature_name=self.feature_name,
1417
1418
                      categorical_feature=self.categorical_feature, params=params,
                      free_raw_data=self.free_raw_data)
wxchan's avatar
wxchan committed
1419
        ret._predictor = self._predictor
1420
        ret.pandas_categorical = self.pandas_categorical
1421
        ret.used_indices = sorted(used_indices)
wxchan's avatar
wxchan committed
1422
1423
1424
        return ret

    def save_binary(self, filename):
1425
        """Save Dataset to a binary file.
wxchan's avatar
wxchan committed
1426

1427
1428
1429
1430
1431
        .. 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
1432
1433
1434
1435
        Parameters
        ----------
        filename : string
            Name of the output file.
Nikita Titov's avatar
Nikita Titov committed
1436
1437
1438
1439
1440

        Returns
        -------
        self : Dataset
            Returns self.
wxchan's avatar
wxchan committed
1441
1442
1443
1444
        """
        _safe_call(_LIB.LGBM_DatasetSaveBinary(
            self.construct().handle,
            c_str(filename)))
Nikita Titov's avatar
Nikita Titov committed
1445
        return self
wxchan's avatar
wxchan committed
1446
1447

    def _update_params(self, params):
1448
1449
        if not params:
            return self
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
        params = copy.deepcopy(params)

        def update():
            if not self.params:
                self.params = params
            else:
                self.params_back_up = copy.deepcopy(self.params)
                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:
1471
                    raise LightGBMError(_LIB.LGBM_GetLastError().decode('utf-8'))
Nikita Titov's avatar
Nikita Titov committed
1472
        return self
wxchan's avatar
wxchan committed
1473

1474
    def _reverse_update_params(self):
1475
1476
1477
        if self.handle is None:
            self.params = copy.deepcopy(self.params_back_up)
            self.params_back_up = None
Nikita Titov's avatar
Nikita Titov committed
1478
        return self
1479

wxchan's avatar
wxchan committed
1480
    def set_field(self, field_name, data):
wxchan's avatar
wxchan committed
1481
        """Set property into the Dataset.
wxchan's avatar
wxchan committed
1482
1483
1484

        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
1485
        field_name : string
1486
            The field name of the information.
1487
        data : list, numpy 1-D array, pandas Series or None
1488
            The array of data to be set.
Nikita Titov's avatar
Nikita Titov committed
1489
1490
1491
1492
1493

        Returns
        -------
        self : Dataset
            Dataset with set property.
wxchan's avatar
wxchan committed
1494
        """
1495
1496
        if self.handle is None:
            raise Exception("Cannot set %s before construct dataset" % field_name)
wxchan's avatar
wxchan committed
1497
        if data is None:
1498
            # set to None
wxchan's avatar
wxchan committed
1499
1500
1501
1502
            _safe_call(_LIB.LGBM_DatasetSetField(
                self.handle,
                c_str(field_name),
                None,
Guolin Ke's avatar
Guolin Ke committed
1503
1504
                ctypes.c_int(0),
                ctypes.c_int(FIELD_TYPE_MAPPER[field_name])))
Nikita Titov's avatar
Nikita Titov committed
1505
            return self
Guolin Ke's avatar
Guolin Ke committed
1506
1507
1508
1509
1510
        dtype = np.float32
        if field_name == 'group':
            dtype = np.int32
        elif field_name == 'init_score':
            dtype = np.float64
1511
        data = list_to_1d_numpy(data, dtype, name=field_name)
1512
1513
        if data.dtype == np.float32 or data.dtype == np.float64:
            ptr_data, type_data, _ = c_float_array(data)
wxchan's avatar
wxchan committed
1514
        elif data.dtype == np.int32:
1515
            ptr_data, type_data, _ = c_int_array(data)
wxchan's avatar
wxchan committed
1516
        else:
Nikita Titov's avatar
Nikita Titov committed
1517
            raise TypeError("Expected np.float32/64 or np.int32, met type({})".format(data.dtype))
wxchan's avatar
wxchan committed
1518
        if type_data != FIELD_TYPE_MAPPER[field_name]:
1519
            raise TypeError("Input type error for set_field")
wxchan's avatar
wxchan committed
1520
1521
1522
1523
        _safe_call(_LIB.LGBM_DatasetSetField(
            self.handle,
            c_str(field_name),
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
1524
1525
            ctypes.c_int(len(data)),
            ctypes.c_int(type_data)))
1526
        self.version += 1
Nikita Titov's avatar
Nikita Titov committed
1527
        return self
wxchan's avatar
wxchan committed
1528

wxchan's avatar
wxchan committed
1529
1530
    def get_field(self, field_name):
        """Get property from the Dataset.
wxchan's avatar
wxchan committed
1531
1532
1533

        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
1534
        field_name : string
1535
            The field name of the information.
wxchan's avatar
wxchan committed
1536
1537
1538

        Returns
        -------
1539
1540
        info : numpy array
            A numpy array with information from the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1541
        """
1542
        if self.handle is None:
1543
            raise Exception("Cannot get %s before construct Dataset" % field_name)
Guolin Ke's avatar
Guolin Ke committed
1544
1545
        tmp_out_len = ctypes.c_int()
        out_type = ctypes.c_int()
wxchan's avatar
wxchan committed
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
        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:
            return cint32_array_to_numpy(ctypes.cast(ret, ctypes.POINTER(ctypes.c_int32)), tmp_out_len.value)
        elif out_type.value == C_API_DTYPE_FLOAT32:
            return cfloat32_array_to_numpy(ctypes.cast(ret, ctypes.POINTER(ctypes.c_float)), tmp_out_len.value)
Guolin Ke's avatar
Guolin Ke committed
1561
1562
        elif out_type.value == C_API_DTYPE_FLOAT64:
            return cfloat64_array_to_numpy(ctypes.cast(ret, ctypes.POINTER(ctypes.c_double)), tmp_out_len.value)
1563
        else:
wxchan's avatar
wxchan committed
1564
            raise TypeError("Unknown type")
Guolin Ke's avatar
Guolin Ke committed
1565

1566
    def set_categorical_feature(self, categorical_feature):
1567
        """Set categorical features.
1568
1569
1570

        Parameters
        ----------
1571
1572
        categorical_feature : list of int or strings
            Names or indices of categorical features.
Nikita Titov's avatar
Nikita Titov committed
1573
1574
1575
1576
1577

        Returns
        -------
        self : Dataset
            Dataset with set categorical features.
1578
1579
        """
        if self.categorical_feature == categorical_feature:
Nikita Titov's avatar
Nikita Titov committed
1580
            return self
1581
        if self.data is not None:
1582
1583
            if self.categorical_feature is None:
                self.categorical_feature = categorical_feature
Nikita Titov's avatar
Nikita Titov committed
1584
                return self._free_handle()
1585
1586
            elif categorical_feature == 'auto':
                warnings.warn('Using categorical_feature in Dataset.')
Nikita Titov's avatar
Nikita Titov committed
1587
                return self
1588
            else:
1589
1590
                warnings.warn('categorical_feature in Dataset is overridden.\n'
                              'New categorical_feature is {}'.format(sorted(list(categorical_feature))))
1591
                self.categorical_feature = categorical_feature
Nikita Titov's avatar
Nikita Titov committed
1592
                return self._free_handle()
1593
        else:
1594
1595
            raise LightGBMError("Cannot set categorical feature after freed raw data, "
                                "set free_raw_data=False when construct Dataset to avoid this.")
1596

Guolin Ke's avatar
Guolin Ke committed
1597
    def _set_predictor(self, predictor):
1598
1599
1600
1601
        """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
1602
        """
1603
        if predictor is self._predictor and (predictor is None or predictor.current_iteration() == self._predictor.current_iteration()):
Nikita Titov's avatar
Nikita Titov committed
1604
            return self
1605
        if self.handle is None:
Guolin Ke's avatar
Guolin Ke committed
1606
            self._predictor = predictor
1607
1608
1609
1610
1611
1612
        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
1613
        else:
1614
1615
            raise LightGBMError("Cannot set predictor after freed raw data, "
                                "set free_raw_data=False when construct Dataset to avoid this.")
1616
        return self
Guolin Ke's avatar
Guolin Ke committed
1617
1618

    def set_reference(self, reference):
1619
        """Set reference Dataset.
Guolin Ke's avatar
Guolin Ke committed
1620
1621
1622
1623

        Parameters
        ----------
        reference : Dataset
1624
            Reference that is used as a template to construct the current Dataset.
Nikita Titov's avatar
Nikita Titov committed
1625
1626
1627
1628
1629

        Returns
        -------
        self : Dataset
            Dataset with set reference.
Guolin Ke's avatar
Guolin Ke committed
1630
        """
1631
1632
1633
        self.set_categorical_feature(reference.categorical_feature) \
            .set_feature_name(reference.feature_name) \
            ._set_predictor(reference._predictor)
1634
1635
        # we're done if self and reference share a common upstrem reference
        if self.get_ref_chain().intersection(reference.get_ref_chain()):
Nikita Titov's avatar
Nikita Titov committed
1636
            return self
Guolin Ke's avatar
Guolin Ke committed
1637
1638
        if self.data is not None:
            self.reference = reference
Nikita Titov's avatar
Nikita Titov committed
1639
            return self._free_handle()
Guolin Ke's avatar
Guolin Ke committed
1640
        else:
1641
1642
            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
1643
1644

    def set_feature_name(self, feature_name):
1645
        """Set feature name.
Guolin Ke's avatar
Guolin Ke committed
1646
1647
1648

        Parameters
        ----------
1649
1650
        feature_name : list of strings
            Feature names.
Nikita Titov's avatar
Nikita Titov committed
1651
1652
1653
1654
1655

        Returns
        -------
        self : Dataset
            Dataset with set feature name.
Guolin Ke's avatar
Guolin Ke committed
1656
        """
1657
1658
        if feature_name != 'auto':
            self.feature_name = feature_name
1659
        if self.handle is not None and feature_name is not None and feature_name != 'auto':
wxchan's avatar
wxchan committed
1660
            if len(feature_name) != self.num_feature():
1661
1662
                raise ValueError("Length of feature_name({}) and num_feature({}) don't match"
                                 .format(len(feature_name), self.num_feature()))
1663
            c_feature_name = [c_str(name) for name in feature_name]
wxchan's avatar
wxchan committed
1664
1665
1666
            _safe_call(_LIB.LGBM_DatasetSetFeatureNames(
                self.handle,
                c_array(ctypes.c_char_p, c_feature_name),
Guolin Ke's avatar
Guolin Ke committed
1667
                ctypes.c_int(len(feature_name))))
Nikita Titov's avatar
Nikita Titov committed
1668
        return self
Guolin Ke's avatar
Guolin Ke committed
1669
1670

    def set_label(self, label):
1671
        """Set label of Dataset.
Guolin Ke's avatar
Guolin Ke committed
1672
1673
1674

        Parameters
        ----------
1675
        label : list, numpy 1-D array, pandas Series / one-column DataFrame or None
1676
            The label information to be set into Dataset.
Nikita Titov's avatar
Nikita Titov committed
1677
1678
1679
1680
1681

        Returns
        -------
        self : Dataset
            Dataset with set label.
Guolin Ke's avatar
Guolin Ke committed
1682
1683
        """
        self.label = label
1684
        if self.handle is not None:
1685
            label = list_to_1d_numpy(_label_from_pandas(label), name='label')
wxchan's avatar
wxchan committed
1686
            self.set_field('label', label)
1687
            self.label = self.get_field('label')  # original values can be modified at cpp side
Nikita Titov's avatar
Nikita Titov committed
1688
        return self
Guolin Ke's avatar
Guolin Ke committed
1689
1690

    def set_weight(self, weight):
1691
        """Set weight of each instance.
Guolin Ke's avatar
Guolin Ke committed
1692
1693
1694

        Parameters
        ----------
1695
        weight : list, numpy 1-D array, pandas Series or None
1696
            Weight to be set for each data point.
Nikita Titov's avatar
Nikita Titov committed
1697
1698
1699
1700
1701

        Returns
        -------
        self : Dataset
            Dataset with set weight.
Guolin Ke's avatar
Guolin Ke committed
1702
        """
1703
1704
        if weight is not None and np.all(weight == 1):
            weight = None
Guolin Ke's avatar
Guolin Ke committed
1705
        self.weight = weight
1706
        if self.handle is not None and weight is not None:
wxchan's avatar
wxchan committed
1707
1708
            weight = list_to_1d_numpy(weight, name='weight')
            self.set_field('weight', weight)
1709
            self.weight = self.get_field('weight')  # original values can be modified at cpp side
Nikita Titov's avatar
Nikita Titov committed
1710
        return self
Guolin Ke's avatar
Guolin Ke committed
1711
1712

    def set_init_score(self, init_score):
1713
        """Set init score of Booster to start from.
Guolin Ke's avatar
Guolin Ke committed
1714
1715
1716

        Parameters
        ----------
1717
        init_score : list, numpy 1-D array, pandas Series or None
1718
            Init score for Booster.
Nikita Titov's avatar
Nikita Titov committed
1719
1720
1721
1722
1723

        Returns
        -------
        self : Dataset
            Dataset with set init score.
Guolin Ke's avatar
Guolin Ke committed
1724
1725
        """
        self.init_score = init_score
1726
        if self.handle is not None and init_score is not None:
Guolin Ke's avatar
Guolin Ke committed
1727
            init_score = list_to_1d_numpy(init_score, np.float64, name='init_score')
wxchan's avatar
wxchan committed
1728
            self.set_field('init_score', init_score)
1729
            self.init_score = self.get_field('init_score')  # original values can be modified at cpp side
Nikita Titov's avatar
Nikita Titov committed
1730
        return self
Guolin Ke's avatar
Guolin Ke committed
1731
1732

    def set_group(self, group):
1733
        """Set group size of Dataset (used for ranking).
Guolin Ke's avatar
Guolin Ke committed
1734
1735
1736

        Parameters
        ----------
1737
        group : list, numpy 1-D array, pandas Series or None
1738
1739
1740
            Group/query data.
            Only used in the learning-to-rank task.
            sum(group) = n_samples.
1741
1742
            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
1743
1744
1745
1746
1747

        Returns
        -------
        self : Dataset
            Dataset with set group.
Guolin Ke's avatar
Guolin Ke committed
1748
1749
        """
        self.group = group
1750
        if self.handle is not None and group is not None:
wxchan's avatar
wxchan committed
1751
1752
            group = list_to_1d_numpy(group, np.int32, name='group')
            self.set_field('group', group)
Nikita Titov's avatar
Nikita Titov committed
1753
        return self
Guolin Ke's avatar
Guolin Ke committed
1754

1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
    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)
1769
        string_buffers = [ctypes.create_string_buffer(reserved_string_buffer_size) for i in range(num_feature)]
1770
1771
1772
        ptr_string_buffers = (ctypes.c_char_p * num_feature)(*map(ctypes.addressof, string_buffers))
        _safe_call(_LIB.LGBM_DatasetGetFeatureNames(
            self.handle,
1773
            ctypes.c_int(num_feature),
1774
            ctypes.byref(tmp_out_len),
1775
            ctypes.c_size_t(reserved_string_buffer_size),
1776
1777
1778
1779
1780
1781
1782
1783
1784
            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")
        if reserved_string_buffer_size < required_string_buffer_size.value:
            raise BufferError(
                "Allocated feature name buffer size ({}) was inferior to the needed size ({})."
                .format(reserved_string_buffer_size, required_string_buffer_size.value)
            )
1785
        return [string_buffers[i].value.decode('utf-8') for i in range(num_feature)]
1786

Guolin Ke's avatar
Guolin Ke committed
1787
    def get_label(self):
1788
        """Get the label of the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1789
1790
1791

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1792
        label : numpy array or None
1793
            The label information from the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1794
        """
1795
        if self.label is None:
wxchan's avatar
wxchan committed
1796
            self.label = self.get_field('label')
Guolin Ke's avatar
Guolin Ke committed
1797
1798
1799
        return self.label

    def get_weight(self):
1800
        """Get the weight of the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1801
1802
1803

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1804
        weight : numpy array or None
1805
            Weight for each data point from the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1806
        """
1807
        if self.weight is None:
wxchan's avatar
wxchan committed
1808
            self.weight = self.get_field('weight')
Guolin Ke's avatar
Guolin Ke committed
1809
1810
1811
        return self.weight

    def get_init_score(self):
1812
        """Get the initial score of the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1813
1814
1815

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1816
        init_score : numpy array or None
1817
            Init score of Booster.
Guolin Ke's avatar
Guolin Ke committed
1818
        """
1819
        if self.init_score is None:
wxchan's avatar
wxchan committed
1820
            self.init_score = self.get_field('init_score')
Guolin Ke's avatar
Guolin Ke committed
1821
1822
        return self.init_score

1823
1824
1825
1826
1827
    def get_data(self):
        """Get the raw data of the Dataset.

        Returns
        -------
1828
        data : string, numpy array, pandas DataFrame, H2O DataTable's Frame, scipy.sparse, list of numpy arrays or None
1829
1830
1831
1832
            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
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
        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, :]
                elif isinstance(self.data, DataFrame):
                    self.data = self.data.iloc[self.used_indices].copy()
                elif isinstance(self.data, DataTable):
                    self.data = self.data[self.used_indices, :]
                else:
                    warnings.warn("Cannot subset {} type of raw data.\n"
                                  "Returning original raw data".format(type(self.data).__name__))
1845
            self.need_slice = False
Guolin Ke's avatar
Guolin Ke committed
1846
1847
1848
        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.")
1849
1850
        return self.data

Guolin Ke's avatar
Guolin Ke committed
1851
    def get_group(self):
1852
        """Get the group of the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1853
1854
1855

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1856
        group : numpy array or None
1857
1858
1859
            Group/query data.
            Only used in the learning-to-rank task.
            sum(group) = n_samples.
1860
1861
            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
1862
        """
1863
        if self.group is None:
wxchan's avatar
wxchan committed
1864
            self.group = self.get_field('group')
Guolin Ke's avatar
Guolin Ke committed
1865
1866
            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
1867
                self.group = np.diff(self.group)
Guolin Ke's avatar
Guolin Ke committed
1868
1869
1870
        return self.group

    def num_data(self):
1871
        """Get the number of rows in the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1872
1873
1874

        Returns
        -------
1875
1876
        number_of_rows : int
            The number of rows in the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1877
        """
1878
        if self.handle is not None:
Guolin Ke's avatar
Guolin Ke committed
1879
            ret = ctypes.c_int()
wxchan's avatar
wxchan committed
1880
1881
1882
            _safe_call(_LIB.LGBM_DatasetGetNumData(self.handle,
                                                   ctypes.byref(ret)))
            return ret.value
Guolin Ke's avatar
Guolin Ke committed
1883
        else:
1884
            raise LightGBMError("Cannot get num_data before construct dataset")
Guolin Ke's avatar
Guolin Ke committed
1885
1886

    def num_feature(self):
1887
        """Get the number of columns (features) in the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1888
1889
1890

        Returns
        -------
1891
1892
        number_of_columns : int
            The number of columns (features) in the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1893
        """
1894
        if self.handle is not None:
Guolin Ke's avatar
Guolin Ke committed
1895
            ret = ctypes.c_int()
wxchan's avatar
wxchan committed
1896
1897
1898
            _safe_call(_LIB.LGBM_DatasetGetNumFeature(self.handle,
                                                      ctypes.byref(ret)))
            return ret.value
Guolin Ke's avatar
Guolin Ke committed
1899
        else:
1900
            raise LightGBMError("Cannot get num_feature before construct dataset")
Guolin Ke's avatar
Guolin Ke committed
1901

1902
    def get_ref_chain(self, ref_limit=100):
1903
1904
1905
1906
1907
        """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.
1908
1909
1910
1911
1912

        Parameters
        ----------
        ref_limit : int, optional (default=100)
            The limit number of references.
1913
1914
1915

        Returns
        -------
1916
1917
1918
        ref_chain : set of Dataset
            Chain of references of the Datasets.
        """
1919
        head = self
1920
        ref_chain = set()
1921
1922
        while len(ref_chain) < ref_limit:
            if isinstance(head, Dataset):
1923
                ref_chain.add(head)
1924
1925
1926
1927
1928
1929
                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
1930
        return ref_chain
1931

1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
    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
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
        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()))
                elif isinstance(other.data, DataFrame):
                    self.data = np.hstack((self.data, other.data.values))
                elif isinstance(other.data, DataTable):
                    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)
                elif isinstance(other.data, DataFrame):
                    self.data = scipy.sparse.hstack((self.data, other.data.values), format=sparse_format)
                elif isinstance(other.data, DataTable):
                    self.data = scipy.sparse.hstack((self.data, other.data.to_numpy()), format=sparse_format)
                else:
                    self.data = None
            elif isinstance(self.data, DataFrame):
                if not PANDAS_INSTALLED:
                    raise LightGBMError("Cannot add features to DataFrame type of raw data "
                                        "without pandas installed")
                from pandas import concat
                if isinstance(other.data, np.ndarray):
                    self.data = concat((self.data, DataFrame(other.data)),
                                       axis=1, ignore_index=True)
                elif scipy.sparse.issparse(other.data):
                    self.data = concat((self.data, DataFrame(other.data.toarray())),
                                       axis=1, ignore_index=True)
                elif isinstance(other.data, DataFrame):
                    self.data = concat((self.data, other.data),
                                       axis=1, ignore_index=True)
                elif isinstance(other.data, DataTable):
                    self.data = concat((self.data, DataFrame(other.data.to_numpy())),
                                       axis=1, ignore_index=True)
                else:
                    self.data = None
            elif isinstance(self.data, DataTable):
                if isinstance(other.data, np.ndarray):
                    self.data = DataTable(np.hstack((self.data.to_numpy(), other.data)))
                elif scipy.sparse.issparse(other.data):
                    self.data = DataTable(np.hstack((self.data.to_numpy(), other.data.toarray())))
                elif isinstance(other.data, DataFrame):
                    self.data = DataTable(np.hstack((self.data.to_numpy(), other.data.values)))
                elif isinstance(other.data, DataTable):
                    self.data = DataTable(np.hstack((self.data.to_numpy(), other.data.to_numpy())))
                else:
                    self.data = None
            else:
                self.data = None
        if self.data is None:
            err_msg = ("Cannot add features from {} type of raw data to "
                       "{} type of raw data.\n").format(type(other.data).__name__,
                                                        old_self_data_type)
            err_msg += ("Set free_raw_data=False when construct Dataset to avoid this"
                        if was_none else "Freeing raw data")
            warnings.warn(err_msg)
        self.feature_name = self.get_feature_name()
        warnings.warn("Reseting categorical features.\n"
                      "You can set new categorical features via ``set_categorical_feature`` method")
        self.categorical_feature = "auto"
        self.pandas_categorical = None
2020
2021
        return self

2022
    def _dump_text(self, filename):
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
        """Save Dataset to a text file.

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

        Parameters
        ----------
        filename : string
            Name of the output file.

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

wxchan's avatar
wxchan committed
2042

2043
class Booster:
2044
    """Booster in LightGBM."""
2045

2046
    def __init__(self, params=None, train_set=None, model_file=None, model_str=None, silent=False):
2047
        """Initialize the Booster.
wxchan's avatar
wxchan committed
2048
2049
2050

        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
2051
        params : dict or None, optional (default=None)
2052
2053
2054
2055
            Parameters for Booster.
        train_set : Dataset or None, optional (default=None)
            Training dataset.
        model_file : string or None, optional (default=None)
wxchan's avatar
wxchan committed
2056
            Path to the model file.
2057
2058
        model_str : string or None, optional (default=None)
            Model will be loaded from this string.
2059
2060
        silent : bool, optional (default=False)
            Whether to print messages during construction.
wxchan's avatar
wxchan committed
2061
        """
2062
        self.handle = None
2063
        self.network = False
wxchan's avatar
wxchan committed
2064
        self.__need_reload_eval_info = True
2065
        self._train_data_name = "training"
wxchan's avatar
wxchan committed
2066
        self.__attr = {}
2067
        self.__set_objective_to_none = False
wxchan's avatar
wxchan committed
2068
        self.best_iteration = -1
wxchan's avatar
wxchan committed
2069
        self.best_score = {}
2070
        params = {} if params is None else copy.deepcopy(params)
2071
        # user can set verbose with params, it has higher priority
2072
        if not any(verbose_alias in params for verbose_alias in _ConfigAliases.get("verbosity")) and silent:
2073
            params["verbose"] = -1
wxchan's avatar
wxchan committed
2074
        if train_set is not None:
2075
            # Training task
wxchan's avatar
wxchan committed
2076
            if not isinstance(train_set, Dataset):
2077
2078
                raise TypeError('Training data should be Dataset instance, met {}'
                                .format(type(train_set).__name__))
2079
            # set network if necessary
2080
            for alias in _ConfigAliases.get("machines"):
2081
2082
                if alias in params:
                    machines = params[alias]
2083
                    if isinstance(machines, str):
2084
2085
2086
2087
2088
2089
2090
2091
2092
                        num_machines = len(machines.split(','))
                    elif isinstance(machines, (list, set)):
                        num_machines = len(machines)
                        machines = ','.join(machines)
                    else:
                        raise ValueError("Invalid machines in params.")
                    self.set_network(machines,
                                     local_listen_port=params.get("local_listen_port", 12400),
                                     listen_time_out=params.get("listen_time_out", 120),
2093
                                     num_machines=params.setdefault("num_machines", num_machines))
2094
                    break
2095
            # construct booster object
2096
2097
2098
2099
            train_set.construct()
            # copy the parameters from train_set
            params.update(train_set.get_params())
            params_str = param_dict_to_str(params)
2100
            self.handle = ctypes.c_void_p()
wxchan's avatar
wxchan committed
2101
            _safe_call(_LIB.LGBM_BoosterCreate(
2102
                train_set.handle,
wxchan's avatar
wxchan committed
2103
2104
                c_str(params_str),
                ctypes.byref(self.handle)))
2105
            # save reference to data
wxchan's avatar
wxchan committed
2106
2107
2108
2109
            self.train_set = train_set
            self.valid_sets = []
            self.name_valid_sets = []
            self.__num_dataset = 1
Guolin Ke's avatar
Guolin Ke committed
2110
2111
            self.__init_predictor = train_set._predictor
            if self.__init_predictor is not None:
wxchan's avatar
wxchan committed
2112
2113
                _safe_call(_LIB.LGBM_BoosterMerge(
                    self.handle,
Guolin Ke's avatar
Guolin Ke committed
2114
                    self.__init_predictor.handle))
Guolin Ke's avatar
Guolin Ke committed
2115
            out_num_class = ctypes.c_int(0)
wxchan's avatar
wxchan committed
2116
2117
2118
2119
            _safe_call(_LIB.LGBM_BoosterGetNumClasses(
                self.handle,
                ctypes.byref(out_num_class)))
            self.__num_class = out_num_class.value
2120
            # buffer for inner predict
wxchan's avatar
wxchan committed
2121
2122
2123
            self.__inner_predict_buffer = [None]
            self.__is_predicted_cur_iter = [False]
            self.__get_eval_info()
2124
            self.pandas_categorical = train_set.pandas_categorical
2125
            self.train_set_version = train_set.version
wxchan's avatar
wxchan committed
2126
        elif model_file is not None:
2127
            # Prediction task
Guolin Ke's avatar
Guolin Ke committed
2128
            out_num_iterations = ctypes.c_int(0)
2129
            self.handle = ctypes.c_void_p()
wxchan's avatar
wxchan committed
2130
2131
2132
2133
            _safe_call(_LIB.LGBM_BoosterCreateFromModelfile(
                c_str(model_file),
                ctypes.byref(out_num_iterations),
                ctypes.byref(self.handle)))
Guolin Ke's avatar
Guolin Ke committed
2134
            out_num_class = ctypes.c_int(0)
wxchan's avatar
wxchan committed
2135
2136
2137
2138
            _safe_call(_LIB.LGBM_BoosterGetNumClasses(
                self.handle,
                ctypes.byref(out_num_class)))
            self.__num_class = out_num_class.value
2139
            self.pandas_categorical = _load_pandas_categorical(file_name=model_file)
2140
2141
        elif model_str is not None:
            self.model_from_string(model_str, not silent)
wxchan's avatar
wxchan committed
2142
        else:
2143
2144
            raise TypeError('Need at least one training dataset or model file or model string '
                            'to create Booster instance')
2145
        self.params = params
wxchan's avatar
wxchan committed
2146
2147

    def __del__(self):
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
        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
2158

wxchan's avatar
wxchan committed
2159
2160
2161
2162
    def __copy__(self):
        return self.__deepcopy__(None)

    def __deepcopy__(self, _):
2163
        model_str = self.model_to_string(num_iteration=-1)
2164
        booster = Booster(model_str=model_str)
2165
        return booster
wxchan's avatar
wxchan committed
2166
2167
2168
2169
2170
2171
2172

    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:
2173
            this["handle"] = self.model_to_string(num_iteration=-1)
wxchan's avatar
wxchan committed
2174
2175
2176
        return this

    def __setstate__(self, state):
2177
2178
        model_str = state.get('handle', None)
        if model_str is not None:
wxchan's avatar
wxchan committed
2179
            handle = ctypes.c_void_p()
Guolin Ke's avatar
Guolin Ke committed
2180
            out_num_iterations = ctypes.c_int(0)
2181
2182
2183
2184
            _safe_call(_LIB.LGBM_BoosterLoadModelFromString(
                c_str(model_str),
                ctypes.byref(out_num_iterations),
                ctypes.byref(handle)))
wxchan's avatar
wxchan committed
2185
2186
2187
            state['handle'] = handle
        self.__dict__.update(state)

wxchan's avatar
wxchan committed
2188
    def free_dataset(self):
Nikita Titov's avatar
Nikita Titov committed
2189
2190
2191
2192
2193
2194
2195
        """Free Booster's Datasets.

        Returns
        -------
        self : Booster
            Booster without Datasets.
        """
wxchan's avatar
wxchan committed
2196
2197
        self.__dict__.pop('train_set', None)
        self.__dict__.pop('valid_sets', None)
2198
        self.__num_dataset = 0
Nikita Titov's avatar
Nikita Titov committed
2199
        return self
wxchan's avatar
wxchan committed
2200

2201
2202
2203
    def _free_buffer(self):
        self.__inner_predict_buffer = []
        self.__is_predicted_cur_iter = []
Nikita Titov's avatar
Nikita Titov committed
2204
        return self
2205

2206
2207
2208
2209
2210
2211
    def set_network(self, machines, local_listen_port=12400,
                    listen_time_out=120, num_machines=1):
        """Set the network configuration.

        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
2212
        machines : list, set or string
2213
            Names of machines.
Nikita Titov's avatar
Nikita Titov committed
2214
        local_listen_port : int, optional (default=12400)
2215
            TCP listen port for local machines.
Nikita Titov's avatar
Nikita Titov committed
2216
        listen_time_out : int, optional (default=120)
2217
            Socket time-out in minutes.
Nikita Titov's avatar
Nikita Titov committed
2218
        num_machines : int, optional (default=1)
2219
            The number of machines for parallel learning application.
Nikita Titov's avatar
Nikita Titov committed
2220
2221
2222
2223
2224

        Returns
        -------
        self : Booster
            Booster with set network.
2225
2226
2227
2228
2229
2230
        """
        _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
2231
        return self
2232
2233

    def free_network(self):
Nikita Titov's avatar
Nikita Titov committed
2234
2235
2236
2237
2238
2239
2240
        """Free Booster's network.

        Returns
        -------
        self : Booster
            Booster with freed network.
        """
2241
2242
        _safe_call(_LIB.LGBM_NetworkFree())
        self.network = False
Nikita Titov's avatar
Nikita Titov committed
2243
        return self
2244

2245
2246
2247
    def trees_to_dataframe(self):
        """Parse the fitted model and return in an easy-to-read pandas DataFrame.

2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
        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.
            - ``node_index`` : string, unique identifier for a node.
            - ``left_child`` : string, ``node_index`` of the child node to the left of a split. ``None`` for leaf nodes.
            - ``right_child`` : string, ``node_index`` of the child node to the right of a split. ``None`` for leaf nodes.
            - ``parent_index`` : string, ``node_index`` of this node's parent. ``None`` for the root node.
            - ``split_feature`` : string, name of the feature used for splitting. ``None`` for leaf nodes.
            - ``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.
2259
2260
2261
            - ``decision_type`` : string, logical operator describing how to compare a value to ``threshold``.
              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.
2262
2263
2264
2265
2266
2267
            - ``missing_direction`` : string, split direction that missing values should go to. ``None`` for leaf nodes.
            - ``missing_type`` : string, describes what types of values are treated as missing.
            - ``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.

2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
        Returns
        -------
        result : pandas DataFrame
            Returns a pandas DataFrame of the parsed model.
        """
        if not PANDAS_INSTALLED:
            raise LightGBMError('This method cannot be run without pandas installed')

        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):
                tree_num = str(tree_index) + '-' if tree_index is not None else ''
                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
                node_num = str(tree.get('split_index' if is_split else 'leaf_index', 0))
                return tree_num + node_type + node_num

            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):
2304
                return set(tree.keys()) == {'leaf_value'}
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379

            # 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))

        return DataFrame(model_list, columns=model_list[0].keys())

wxchan's avatar
wxchan committed
2380
    def set_train_data_name(self, name):
2381
2382
2383
2384
        """Set the name to the training Dataset.

        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
2385
2386
2387
2388
2389
2390
2391
        name : string
            Name for the training Dataset.

        Returns
        -------
        self : Booster
            Booster with set training Dataset name.
2392
        """
2393
        self._train_data_name = name
Nikita Titov's avatar
Nikita Titov committed
2394
        return self
wxchan's avatar
wxchan committed
2395
2396

    def add_valid(self, data, name):
2397
        """Add validation data.
wxchan's avatar
wxchan committed
2398
2399
2400
2401

        Parameters
        ----------
        data : Dataset
2402
2403
2404
            Validation data.
        name : string
            Name of validation data.
Nikita Titov's avatar
Nikita Titov committed
2405
2406
2407
2408
2409

        Returns
        -------
        self : Booster
            Booster with set validation data.
wxchan's avatar
wxchan committed
2410
        """
Guolin Ke's avatar
Guolin Ke committed
2411
        if not isinstance(data, Dataset):
2412
2413
            raise TypeError('Validation data should be Dataset instance, met {}'
                            .format(type(data).__name__))
Guolin Ke's avatar
Guolin Ke committed
2414
        if data._predictor is not self.__init_predictor:
2415
2416
            raise LightGBMError("Add validation data failed, "
                                "you should use same predictor for these data")
wxchan's avatar
wxchan committed
2417
2418
        _safe_call(_LIB.LGBM_BoosterAddValidData(
            self.handle,
wxchan's avatar
wxchan committed
2419
            data.construct().handle))
wxchan's avatar
wxchan committed
2420
2421
2422
2423
2424
        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
2425
        return self
wxchan's avatar
wxchan committed
2426
2427

    def reset_parameter(self, params):
2428
        """Reset parameters of Booster.
wxchan's avatar
wxchan committed
2429
2430
2431
2432

        Parameters
        ----------
        params : dict
2433
            New parameters for Booster.
Nikita Titov's avatar
Nikita Titov committed
2434
2435
2436
2437
2438

        Returns
        -------
        self : Booster
            Booster with new parameters.
wxchan's avatar
wxchan committed
2439
2440
2441
2442
2443
2444
        """
        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
2445
        self.params.update(params)
Nikita Titov's avatar
Nikita Titov committed
2446
        return self
wxchan's avatar
wxchan committed
2447
2448

    def update(self, train_set=None, fobj=None):
Nikita Titov's avatar
Nikita Titov committed
2449
        """Update Booster for one iteration.
2450

wxchan's avatar
wxchan committed
2451
2452
        Parameters
        ----------
2453
2454
2455
2456
        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
2457
            Customized objective function.
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
            Should accept two parameters: preds, train_data,
            and return (grad, hess).

                preds : list or numpy 1-D array
                    The predicted values.
                train_data : Dataset
                    The training dataset.
                grad : list or numpy 1-D array
                    The value of the first order derivative (gradient) for each sample point.
                hess : list or numpy 1-D array
                    The value of the second order derivative (Hessian) for each sample point.
wxchan's avatar
wxchan committed
2469

2470
            For binary task, the preds is probability of positive class (or margin in case of specified ``fobj``).
2471
2472
            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]
2473
2474
            and you should group grad and hess in this way as well.

wxchan's avatar
wxchan committed
2475
2476
        Returns
        -------
2477
2478
        is_finished : bool
            Whether the update was successfully finished.
wxchan's avatar
wxchan committed
2479
        """
2480
        # need reset training data
2481
2482
2483
2484
2485
2486
        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
2487
            if not isinstance(train_set, Dataset):
2488
2489
                raise TypeError('Training data should be Dataset instance, met {}'
                                .format(type(train_set).__name__))
Guolin Ke's avatar
Guolin Ke committed
2490
            if train_set._predictor is not self.__init_predictor:
2491
2492
                raise LightGBMError("Replace training data failed, "
                                    "you should use same predictor for these data")
wxchan's avatar
wxchan committed
2493
2494
2495
            self.train_set = train_set
            _safe_call(_LIB.LGBM_BoosterResetTrainingData(
                self.handle,
wxchan's avatar
wxchan committed
2496
                self.train_set.construct().handle))
wxchan's avatar
wxchan committed
2497
            self.__inner_predict_buffer[0] = None
2498
            self.train_set_version = self.train_set.version
wxchan's avatar
wxchan committed
2499
2500
        is_finished = ctypes.c_int(0)
        if fobj is None:
2501
            if self.__set_objective_to_none:
2502
                raise LightGBMError('Cannot update due to null objective function.')
wxchan's avatar
wxchan committed
2503
2504
2505
            _safe_call(_LIB.LGBM_BoosterUpdateOneIter(
                self.handle,
                ctypes.byref(is_finished)))
2506
            self.__is_predicted_cur_iter = [False for _ in range(self.__num_dataset)]
wxchan's avatar
wxchan committed
2507
2508
            return is_finished.value == 1
        else:
2509
            if not self.__set_objective_to_none:
Nikita Titov's avatar
Nikita Titov committed
2510
                self.reset_parameter({"objective": "none"}).__set_objective_to_none = True
wxchan's avatar
wxchan committed
2511
2512
2513
2514
            grad, hess = fobj(self.__inner_predict(0), self.train_set)
            return self.__boost(grad, hess)

    def __boost(self, grad, hess):
2515
        """Boost Booster for one iteration with customized gradient statistics.
Nikita Titov's avatar
Nikita Titov committed
2516

Nikita Titov's avatar
Nikita Titov committed
2517
2518
        .. note::

2519
            For binary task, the score is probability of positive class (or margin in case of custom objective).
Nikita Titov's avatar
Nikita Titov committed
2520
2521
2522
            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.
2523

wxchan's avatar
wxchan committed
2524
2525
        Parameters
        ----------
2526
        grad : list or numpy 1-D array
Nikita Titov's avatar
Nikita Titov committed
2527
            The first order derivative (gradient).
2528
        hess : list or numpy 1-D array
Nikita Titov's avatar
Nikita Titov committed
2529
            The second order derivative (Hessian).
wxchan's avatar
wxchan committed
2530
2531
2532

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
2533
2534
        is_finished : bool
            Whether the boost was successfully finished.
wxchan's avatar
wxchan committed
2535
        """
2536
2537
        grad = list_to_1d_numpy(grad, name='gradient')
        hess = list_to_1d_numpy(hess, name='hessian')
2538
2539
        assert grad.flags.c_contiguous
        assert hess.flags.c_contiguous
wxchan's avatar
wxchan committed
2540
        if len(grad) != len(hess):
2541
2542
            raise ValueError("Lengths of gradient({}) and hessian({}) don't match"
                             .format(len(grad), len(hess)))
wxchan's avatar
wxchan committed
2543
2544
2545
2546
2547
2548
        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)))
2549
        self.__is_predicted_cur_iter = [False for _ in range(self.__num_dataset)]
wxchan's avatar
wxchan committed
2550
2551
2552
        return is_finished.value == 1

    def rollback_one_iter(self):
Nikita Titov's avatar
Nikita Titov committed
2553
2554
2555
2556
2557
2558
2559
        """Rollback one iteration.

        Returns
        -------
        self : Booster
            Booster with rolled back one iteration.
        """
wxchan's avatar
wxchan committed
2560
2561
        _safe_call(_LIB.LGBM_BoosterRollbackOneIter(
            self.handle))
2562
        self.__is_predicted_cur_iter = [False for _ in range(self.__num_dataset)]
Nikita Titov's avatar
Nikita Titov committed
2563
        return self
wxchan's avatar
wxchan committed
2564
2565

    def current_iteration(self):
2566
2567
2568
2569
2570
2571
2572
        """Get the index of the current iteration.

        Returns
        -------
        cur_iter : int
            The index of the current iteration.
        """
Guolin Ke's avatar
Guolin Ke committed
2573
        out_cur_iter = ctypes.c_int(0)
wxchan's avatar
wxchan committed
2574
2575
2576
2577
2578
        _safe_call(_LIB.LGBM_BoosterGetCurrentIteration(
            self.handle,
            ctypes.byref(out_cur_iter)))
        return out_cur_iter.value

2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
    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

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
2634
    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
2635
    def eval(self, data, name, feval=None):
2636
        """Evaluate for data.
wxchan's avatar
wxchan committed
2637
2638
2639

        Parameters
        ----------
2640
2641
2642
2643
2644
        data : Dataset
            Data for the evaluating.
        name : string
            Name of the data.
        feval : callable or None, optional (default=None)
2645
            Customized evaluation function.
2646
            Should accept two parameters: preds, eval_data,
2647
            and return (eval_name, eval_result, is_higher_better) or list of such tuples.
2648
2649
2650
2651
2652
2653

                preds : list or numpy 1-D array
                    The predicted values.
                eval_data : Dataset
                    The evaluation dataset.
                eval_name : string
2654
                    The name of evaluation function (without whitespaces).
2655
2656
2657
2658
2659
                eval_result : float
                    The eval result.
                is_higher_better : bool
                    Is eval result higher better, e.g. AUC is ``is_higher_better``.

2660
            For binary task, the preds is probability of positive class (or margin in case of specified ``fobj``).
2661
2662
            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].
2663

wxchan's avatar
wxchan committed
2664
2665
        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
2666
        result : list
2667
            List with evaluation results.
wxchan's avatar
wxchan committed
2668
        """
Guolin Ke's avatar
Guolin Ke committed
2669
2670
        if not isinstance(data, Dataset):
            raise TypeError("Can only eval for Dataset instance")
wxchan's avatar
wxchan committed
2671
2672
2673
2674
        data_idx = -1
        if data is self.train_set:
            data_idx = 0
        else:
2675
            for i in range(len(self.valid_sets)):
wxchan's avatar
wxchan committed
2676
2677
2678
                if data is self.valid_sets[i]:
                    data_idx = i + 1
                    break
2679
        # need to push new valid data
wxchan's avatar
wxchan committed
2680
2681
2682
2683
2684
2685
2686
        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):
2687
        """Evaluate for training data.
wxchan's avatar
wxchan committed
2688
2689
2690

        Parameters
        ----------
2691
        feval : callable or None, optional (default=None)
2692
            Customized evaluation function.
2693
2694
            Should accept two parameters: preds, train_data,
            and return (eval_name, eval_result, is_higher_better) or list of such tuples.
2695
2696
2697
2698
2699
2700

                preds : list or numpy 1-D array
                    The predicted values.
                train_data : Dataset
                    The training dataset.
                eval_name : string
2701
                    The name of evaluation function (without whitespaces).
2702
2703
2704
2705
2706
                eval_result : float
                    The eval result.
                is_higher_better : bool
                    Is eval result higher better, e.g. AUC is ``is_higher_better``.

2707
            For binary task, the preds is probability of positive class (or margin in case of specified ``fobj``).
2708
2709
            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
2710
2711
2712

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
2713
        result : list
2714
            List with evaluation results.
wxchan's avatar
wxchan committed
2715
        """
2716
        return self.__inner_eval(self._train_data_name, 0, feval)
wxchan's avatar
wxchan committed
2717
2718

    def eval_valid(self, feval=None):
2719
        """Evaluate for validation data.
wxchan's avatar
wxchan committed
2720
2721
2722

        Parameters
        ----------
2723
        feval : callable or None, optional (default=None)
2724
            Customized evaluation function.
2725
            Should accept two parameters: preds, valid_data,
2726
            and return (eval_name, eval_result, is_higher_better) or list of such tuples.
2727
2728
2729
2730
2731
2732

                preds : list or numpy 1-D array
                    The predicted values.
                valid_data : Dataset
                    The validation dataset.
                eval_name : string
2733
                    The name of evaluation function (without whitespaces).
2734
2735
2736
2737
2738
                eval_result : float
                    The eval result.
                is_higher_better : bool
                    Is eval result higher better, e.g. AUC is ``is_higher_better``.

2739
            For binary task, the preds is probability of positive class (or margin in case of specified ``fobj``).
2740
2741
            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
2742
2743
2744

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
2745
        result : list
2746
            List with evaluation results.
wxchan's avatar
wxchan committed
2747
        """
2748
        return [item for i in range(1, self.__num_dataset)
wxchan's avatar
wxchan committed
2749
                for item in self.__inner_eval(self.name_valid_sets[i - 1], i, feval)]
wxchan's avatar
wxchan committed
2750

2751
    def save_model(self, filename, num_iteration=None, start_iteration=0, importance_type='split'):
2752
        """Save Booster to file.
wxchan's avatar
wxchan committed
2753
2754
2755

        Parameters
        ----------
2756
2757
        filename : string
            Filename to save Booster.
2758
2759
2760
2761
        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
2762
        start_iteration : int, optional (default=0)
2763
            Start index of the iteration that should be saved.
2764
2765
2766
2767
        importance_type : string, optional (default="split")
            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
2768
2769
2770
2771
2772

        Returns
        -------
        self : Booster
            Returns self.
wxchan's avatar
wxchan committed
2773
        """
2774
        if num_iteration is None:
2775
            num_iteration = self.best_iteration
2776
        importance_type_int = FEATURE_IMPORTANCE_TYPE_MAPPER[importance_type]
wxchan's avatar
wxchan committed
2777
2778
        _safe_call(_LIB.LGBM_BoosterSaveModel(
            self.handle,
2779
            ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
2780
            ctypes.c_int(num_iteration),
2781
            ctypes.c_int(importance_type_int),
wxchan's avatar
wxchan committed
2782
            c_str(filename)))
2783
        _dump_pandas_categorical(self.pandas_categorical, filename)
Nikita Titov's avatar
Nikita Titov committed
2784
        return self
wxchan's avatar
wxchan committed
2785

2786
    def shuffle_models(self, start_iteration=0, end_iteration=-1):
2787
        """Shuffle models.
Nikita Titov's avatar
Nikita Titov committed
2788

2789
2790
2791
        Parameters
        ----------
        start_iteration : int, optional (default=0)
2792
            The first iteration that will be shuffled.
2793
2794
        end_iteration : int, optional (default=-1)
            The last iteration that will be shuffled.
2795
            If <= 0, means the last available iteration.
2796

Nikita Titov's avatar
Nikita Titov committed
2797
2798
2799
2800
        Returns
        -------
        self : Booster
            Booster with shuffled models.
2801
        """
2802
2803
        _safe_call(_LIB.LGBM_BoosterShuffleModels(
            self.handle,
Guolin Ke's avatar
Guolin Ke committed
2804
2805
            ctypes.c_int(start_iteration),
            ctypes.c_int(end_iteration)))
Nikita Titov's avatar
Nikita Titov committed
2806
        return self
2807
2808
2809
2810
2811
2812

    def model_from_string(self, model_str, verbose=True):
        """Load Booster from a string.

        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
2813
        model_str : string
2814
            Model will be loaded from this string.
Nikita Titov's avatar
Nikita Titov committed
2815
2816
        verbose : bool, optional (default=True)
            Whether to print messages while loading model.
2817
2818
2819

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
2820
        self : Booster
2821
2822
            Loaded Booster object.
        """
2823
2824
2825
2826
        if self.handle is not None:
            _safe_call(_LIB.LGBM_BoosterFree(self.handle))
        self._free_buffer()
        self.handle = ctypes.c_void_p()
2827
2828
2829
2830
2831
2832
2833
2834
2835
        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)))
2836
        if verbose:
Nikita Titov's avatar
Nikita Titov committed
2837
            print('Finished loading model, total used %d iterations' % int(out_num_iterations.value))
2838
        self.__num_class = out_num_class.value
2839
        self.pandas_categorical = _load_pandas_categorical(model_str=model_str)
2840
2841
        return self

2842
    def model_to_string(self, num_iteration=None, start_iteration=0, importance_type='split'):
2843
        """Save Booster to string.
2844

2845
2846
2847
2848
2849
2850
        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
2851
        start_iteration : int, optional (default=0)
2852
            Start index of the iteration that should be saved.
2853
2854
2855
2856
        importance_type : string, optional (default="split")
            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.
2857
2858
2859

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
2860
        str_repr : string
2861
2862
            String representation of Booster.
        """
2863
        if num_iteration is None:
2864
            num_iteration = self.best_iteration
2865
        importance_type_int = FEATURE_IMPORTANCE_TYPE_MAPPER[importance_type]
2866
        buffer_len = 1 << 20
2867
        tmp_out_len = ctypes.c_int64(0)
2868
2869
2870
2871
        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,
2872
            ctypes.c_int(start_iteration),
2873
            ctypes.c_int(num_iteration),
2874
            ctypes.c_int(importance_type_int),
2875
            ctypes.c_int64(buffer_len),
2876
2877
2878
            ctypes.byref(tmp_out_len),
            ptr_string_buffer))
        actual_len = tmp_out_len.value
2879
        # if buffer length is not long enough, re-allocate a buffer
2880
2881
2882
2883
2884
        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,
2885
                ctypes.c_int(start_iteration),
2886
                ctypes.c_int(num_iteration),
2887
                ctypes.c_int(importance_type_int),
2888
                ctypes.c_int64(actual_len),
2889
2890
                ctypes.byref(tmp_out_len),
                ptr_string_buffer))
2891
        ret = string_buffer.value.decode('utf-8')
2892
2893
        ret += _dump_pandas_categorical(self.pandas_categorical)
        return ret
2894

2895
    def dump_model(self, num_iteration=None, start_iteration=0, importance_type='split'):
Nikita Titov's avatar
Nikita Titov committed
2896
        """Dump Booster to JSON format.
wxchan's avatar
wxchan committed
2897

2898
2899
        Parameters
        ----------
2900
2901
2902
2903
        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
2904
        start_iteration : int, optional (default=0)
2905
            Start index of the iteration that should be dumped.
2906
2907
2908
2909
        importance_type : string, optional (default="split")
            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.
2910

wxchan's avatar
wxchan committed
2911
2912
        Returns
        -------
2913
        json_repr : dict
Nikita Titov's avatar
Nikita Titov committed
2914
            JSON format of Booster.
wxchan's avatar
wxchan committed
2915
        """
2916
        if num_iteration is None:
2917
            num_iteration = self.best_iteration
2918
        importance_type_int = FEATURE_IMPORTANCE_TYPE_MAPPER[importance_type]
wxchan's avatar
wxchan committed
2919
        buffer_len = 1 << 20
2920
        tmp_out_len = ctypes.c_int64(0)
wxchan's avatar
wxchan committed
2921
2922
2923
2924
        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,
2925
            ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
2926
            ctypes.c_int(num_iteration),
2927
            ctypes.c_int(importance_type_int),
2928
            ctypes.c_int64(buffer_len),
wxchan's avatar
wxchan committed
2929
            ctypes.byref(tmp_out_len),
Guolin Ke's avatar
Guolin Ke committed
2930
            ptr_string_buffer))
wxchan's avatar
wxchan committed
2931
        actual_len = tmp_out_len.value
2932
        # if buffer length is not long enough, reallocate a buffer
wxchan's avatar
wxchan committed
2933
2934
2935
2936
2937
        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,
2938
                ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
2939
                ctypes.c_int(num_iteration),
2940
                ctypes.c_int(importance_type_int),
2941
                ctypes.c_int64(actual_len),
wxchan's avatar
wxchan committed
2942
                ctypes.byref(tmp_out_len),
Guolin Ke's avatar
Guolin Ke committed
2943
                ptr_string_buffer))
2944
        ret = json.loads(string_buffer.value.decode('utf-8'))
2945
2946
2947
        ret['pandas_categorical'] = json.loads(json.dumps(self.pandas_categorical,
                                                          default=json_default_with_numpy))
        return ret
wxchan's avatar
wxchan committed
2948

2949
    def predict(self, data, start_iteration=0, num_iteration=None,
2950
                raw_score=False, pred_leaf=False, pred_contrib=False,
2951
                data_has_header=False, is_reshape=True, **kwargs):
2952
        """Make a prediction.
wxchan's avatar
wxchan committed
2953
2954
2955

        Parameters
        ----------
2956
        data : string, numpy array, pandas DataFrame, H2O DataTable's Frame or scipy.sparse
2957
2958
            Data source for prediction.
            If string, it represents the path to txt file.
2959
        start_iteration : int, optional (default=0)
2960
            Start index of the iteration to predict.
2961
            If <= 0, starts from the first iteration.
2962
        num_iteration : int or None, optional (default=None)
2963
2964
2965
2966
            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).
2967
2968
2969
2970
        raw_score : bool, optional (default=False)
            Whether to predict raw scores.
        pred_leaf : bool, optional (default=False)
            Whether to predict leaf index.
2971
2972
        pred_contrib : bool, optional (default=False)
            Whether to predict feature contributions.
2973

Nikita Titov's avatar
Nikita Titov committed
2974
2975
2976
2977
2978
2979
2980
            .. 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.
2981

2982
2983
2984
2985
2986
        data_has_header : bool, optional (default=False)
            Whether the data has header.
            Used only if data is string.
        is_reshape : bool, optional (default=True)
            If True, result is reshaped to [nrow, ncol].
2987
2988
        **kwargs
            Other parameters for the prediction.
wxchan's avatar
wxchan committed
2989
2990
2991

        Returns
        -------
2992
        result : numpy array, scipy.sparse or list of scipy.sparse
2993
            Prediction result.
2994
            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
2995
        """
2996
        predictor = self._to_predictor(copy.deepcopy(kwargs))
2997
        if num_iteration is None:
2998
            if start_iteration <= 0:
2999
3000
3001
3002
                num_iteration = self.best_iteration
            else:
                num_iteration = -1
        return predictor.predict(data, start_iteration, num_iteration,
3003
3004
                                 raw_score, pred_leaf, pred_contrib,
                                 data_has_header, is_reshape)
wxchan's avatar
wxchan committed
3005

3006
    def refit(self, data, label, decay_rate=0.9, **kwargs):
Guolin Ke's avatar
Guolin Ke committed
3007
3008
3009
3010
        """Refit the existing Booster by new data.

        Parameters
        ----------
3011
        data : string, numpy array, pandas DataFrame, H2O DataTable's Frame or scipy.sparse
Guolin Ke's avatar
Guolin Ke committed
3012
3013
            Data source for refit.
            If string, it represents the path to txt file.
3014
        label : list, numpy 1-D array or pandas Series / one-column DataFrame
Guolin Ke's avatar
Guolin Ke committed
3015
3016
            Label for refit.
        decay_rate : float, optional (default=0.9)
3017
3018
            Decay rate of refit,
            will use ``leaf_output = decay_rate * old_leaf_output + (1.0 - decay_rate) * new_leaf_output`` to refit trees.
3019
3020
        **kwargs
            Other parameters for refit.
3021
            These parameters will be passed to ``predict`` method.
Guolin Ke's avatar
Guolin Ke committed
3022
3023
3024
3025
3026
3027

        Returns
        -------
        result : Booster
            Refitted Booster.
        """
3028
3029
        if self.__set_objective_to_none:
            raise LightGBMError('Cannot refit due to null objective function.')
3030
        predictor = self._to_predictor(copy.deepcopy(kwargs))
3031
        leaf_preds = predictor.predict(data, -1, pred_leaf=True)
3032
        nrow, ncol = leaf_preds.shape
3033
3034
3035
3036
        out_is_linear = ctypes.c_bool(False)
        _safe_call(_LIB.LGBM_BoosterGetLinear(
            self.handle,
            ctypes.byref(out_is_linear)))
3037
        new_params = copy.deepcopy(self.params)
3038
3039
        new_params["linear_tree"] = out_is_linear.value
        train_set = Dataset(data, label, silent=True, params=new_params)
3040
        new_params['refit_decay_rate'] = decay_rate
3041
        new_booster = Booster(new_params, train_set)
Guolin Ke's avatar
Guolin Ke committed
3042
3043
3044
3045
3046
        # Copy models
        _safe_call(_LIB.LGBM_BoosterMerge(
            new_booster.handle,
            predictor.handle))
        leaf_preds = leaf_preds.reshape(-1)
3047
        ptr_data, _, _ = c_int_array(leaf_preds)
Guolin Ke's avatar
Guolin Ke committed
3048
3049
3050
3051
3052
        _safe_call(_LIB.LGBM_BoosterRefit(
            new_booster.handle,
            ptr_data,
            ctypes.c_int(nrow),
            ctypes.c_int(ncol)))
3053
3054
        new_booster.network = self.network
        new_booster.__attr = self.__attr.copy()
Guolin Ke's avatar
Guolin Ke committed
3055
3056
        return new_booster

3057
    def get_leaf_output(self, tree_id, leaf_id):
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
        """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.
        """
3072
3073
3074
3075
3076
3077
3078
3079
        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

3080
    def _to_predictor(self, pred_parameter=None):
3081
        """Convert to predictor."""
3082
        predictor = _InnerPredictor(booster_handle=self.handle, pred_parameter=pred_parameter)
3083
        predictor.pandas_categorical = self.pandas_categorical
wxchan's avatar
wxchan committed
3084
3085
        return predictor

3086
    def num_feature(self):
3087
3088
3089
3090
3091
3092
3093
        """Get number of features.

        Returns
        -------
        num_feature : int
            The number of features.
        """
3094
3095
3096
3097
3098
3099
        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
3100
    def feature_name(self):
3101
        """Get names of features.
wxchan's avatar
wxchan committed
3102
3103
3104

        Returns
        -------
3105
3106
        result : list
            List with names of features.
wxchan's avatar
wxchan committed
3107
        """
3108
        num_feature = self.num_feature()
3109
        # Get name of features
wxchan's avatar
wxchan committed
3110
        tmp_out_len = ctypes.c_int(0)
3111
3112
        reserved_string_buffer_size = 255
        required_string_buffer_size = ctypes.c_size_t(0)
3113
        string_buffers = [ctypes.create_string_buffer(reserved_string_buffer_size) for i in range(num_feature)]
wxchan's avatar
wxchan committed
3114
3115
3116
        ptr_string_buffers = (ctypes.c_char_p * num_feature)(*map(ctypes.addressof, string_buffers))
        _safe_call(_LIB.LGBM_BoosterGetFeatureNames(
            self.handle,
3117
            ctypes.c_int(num_feature),
wxchan's avatar
wxchan committed
3118
            ctypes.byref(tmp_out_len),
3119
            ctypes.c_size_t(reserved_string_buffer_size),
3120
            ctypes.byref(required_string_buffer_size),
wxchan's avatar
wxchan committed
3121
3122
3123
            ptr_string_buffers))
        if num_feature != tmp_out_len.value:
            raise ValueError("Length of feature names doesn't equal with num_feature")
3124
3125
3126
3127
3128
        if reserved_string_buffer_size < required_string_buffer_size.value:
            raise BufferError(
                "Allocated feature name buffer size ({}) was inferior to the needed size ({})."
                .format(reserved_string_buffer_size, required_string_buffer_size.value)
            )
3129
        return [string_buffers[i].value.decode('utf-8') for i in range(num_feature)]
wxchan's avatar
wxchan committed
3130

3131
    def feature_importance(self, importance_type='split', iteration=None):
3132
        """Get feature importances.
3133

3134
3135
        Parameters
        ----------
3136
3137
3138
3139
        importance_type : string, optional (default="split")
            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.
3140
3141
3142
3143
        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).
3144

3145
3146
        Returns
        -------
3147
3148
        result : numpy array
            Array with feature importances.
3149
        """
3150
3151
        if iteration is None:
            iteration = self.best_iteration
3152
        importance_type_int = FEATURE_IMPORTANCE_TYPE_MAPPER[importance_type]
Nikita Titov's avatar
Nikita Titov committed
3153
        result = np.zeros(self.num_feature(), dtype=np.float64)
3154
3155
3156
3157
3158
3159
        _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:
3160
            return result.astype(np.int32)
3161
3162
        else:
            return result
3163

3164
3165
3166
3167
3168
3169
3170
3171
3172
    def get_split_value_histogram(self, feature, bins=None, xgboost_style=False):
        """Get split value histogram for the specified feature.

        Parameters
        ----------
        feature : int or string
            The feature name or index the histogram is calculated for.
            If int, interpreted as index.
            If string, interpreted as name.
3173

Nikita Titov's avatar
Nikita Titov committed
3174
3175
3176
            .. warning::

                Categorical features are not supported.
3177

3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
        bins : int, string or None, optional (default=None)
            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.
            If string, it should be one from the list of the supported values by ``numpy.histogram()`` function.
        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
3200
                if feature_names is not None and isinstance(feature, str):
3201
3202
3203
3204
                    split_feature = feature_names[root['split_feature']]
                else:
                    split_feature = root['split_feature']
                if split_feature == feature:
3205
                    if isinstance(root['threshold'], str):
3206
3207
3208
                        raise LightGBMError('Cannot compute split value histogram for the categorical feature')
                    else:
                        values.append(root['threshold'])
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
                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'])

3219
        if bins is None or isinstance(bins, int) and xgboost_style:
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
            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:
                return DataFrame(ret, columns=['SplitValue', 'Count'])
            else:
                return ret
        else:
            return hist, bin_edges

wxchan's avatar
wxchan committed
3233
    def __inner_eval(self, data_name, data_idx, feval=None):
3234
        """Evaluate training or validation data."""
wxchan's avatar
wxchan committed
3235
        if data_idx >= self.__num_dataset:
3236
            raise ValueError("Data_idx should be smaller than number of dataset")
wxchan's avatar
wxchan committed
3237
3238
3239
        self.__get_eval_info()
        ret = []
        if self.__num_inner_eval > 0:
3240
            result = np.zeros(self.__num_inner_eval, dtype=np.float64)
Guolin Ke's avatar
Guolin Ke committed
3241
            tmp_out_len = ctypes.c_int(0)
wxchan's avatar
wxchan committed
3242
3243
            _safe_call(_LIB.LGBM_BoosterGetEval(
                self.handle,
Guolin Ke's avatar
Guolin Ke committed
3244
                ctypes.c_int(data_idx),
wxchan's avatar
wxchan committed
3245
                ctypes.byref(tmp_out_len),
Guolin Ke's avatar
Guolin Ke committed
3246
                result.ctypes.data_as(ctypes.POINTER(ctypes.c_double))))
wxchan's avatar
wxchan committed
3247
            if tmp_out_len.value != self.__num_inner_eval:
3248
                raise ValueError("Wrong length of eval results")
3249
            for i in range(self.__num_inner_eval):
3250
3251
                ret.append((data_name, self.__name_inner_eval[i],
                            result[i], self.__higher_better_inner_eval[i]))
3252
3253
        if callable(feval):
            feval = [feval]
wxchan's avatar
wxchan committed
3254
3255
3256
3257
3258
        if feval is not None:
            if data_idx == 0:
                cur_data = self.train_set
            else:
                cur_data = self.valid_sets[data_idx - 1]
3259
3260
3261
3262
3263
3264
3265
3266
3267
            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
3268
3269
3270
3271
                    ret.append((data_name, eval_name, val, is_higher_better))
        return ret

    def __inner_predict(self, data_idx):
3272
        """Predict for training and validation dataset."""
wxchan's avatar
wxchan committed
3273
        if data_idx >= self.__num_dataset:
3274
            raise ValueError("Data_idx should be smaller than number of dataset")
wxchan's avatar
wxchan committed
3275
3276
3277
3278
3279
        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
3280
            self.__inner_predict_buffer[data_idx] = np.zeros(n_preds, dtype=np.float64)
3281
        # avoid to predict many time in one iteration
wxchan's avatar
wxchan committed
3282
3283
        if not self.__is_predicted_cur_iter[data_idx]:
            tmp_out_len = ctypes.c_int64(0)
Guolin Ke's avatar
Guolin Ke committed
3284
            data_ptr = self.__inner_predict_buffer[data_idx].ctypes.data_as(ctypes.POINTER(ctypes.c_double))
wxchan's avatar
wxchan committed
3285
3286
            _safe_call(_LIB.LGBM_BoosterGetPredict(
                self.handle,
Guolin Ke's avatar
Guolin Ke committed
3287
                ctypes.c_int(data_idx),
wxchan's avatar
wxchan committed
3288
3289
3290
                ctypes.byref(tmp_out_len),
                data_ptr))
            if tmp_out_len.value != len(self.__inner_predict_buffer[data_idx]):
3291
                raise ValueError("Wrong length of predict results for data %d" % (data_idx))
wxchan's avatar
wxchan committed
3292
3293
3294
3295
            self.__is_predicted_cur_iter[data_idx] = True
        return self.__inner_predict_buffer[data_idx]

    def __get_eval_info(self):
3296
        """Get inner evaluation count and names."""
wxchan's avatar
wxchan committed
3297
3298
        if self.__need_reload_eval_info:
            self.__need_reload_eval_info = False
Guolin Ke's avatar
Guolin Ke committed
3299
            out_num_eval = ctypes.c_int(0)
3300
            # Get num of inner evals
wxchan's avatar
wxchan committed
3301
3302
3303
3304
3305
            _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:
3306
                # Get name of evals
Guolin Ke's avatar
Guolin Ke committed
3307
                tmp_out_len = ctypes.c_int(0)
3308
3309
3310
                reserved_string_buffer_size = 255
                required_string_buffer_size = ctypes.c_size_t(0)
                string_buffers = [
3311
                    ctypes.create_string_buffer(reserved_string_buffer_size) for i in range(self.__num_inner_eval)
3312
                ]
wxchan's avatar
wxchan committed
3313
                ptr_string_buffers = (ctypes.c_char_p * self.__num_inner_eval)(*map(ctypes.addressof, string_buffers))
wxchan's avatar
wxchan committed
3314
3315
                _safe_call(_LIB.LGBM_BoosterGetEvalNames(
                    self.handle,
3316
                    ctypes.c_int(self.__num_inner_eval),
wxchan's avatar
wxchan committed
3317
                    ctypes.byref(tmp_out_len),
3318
                    ctypes.c_size_t(reserved_string_buffer_size),
3319
                    ctypes.byref(required_string_buffer_size),
wxchan's avatar
wxchan committed
3320
3321
                    ptr_string_buffers))
                if self.__num_inner_eval != tmp_out_len.value:
3322
                    raise ValueError("Length of eval names doesn't equal with num_evals")
3323
3324
3325
3326
3327
                if reserved_string_buffer_size < required_string_buffer_size.value:
                    raise BufferError(
                        "Allocated eval name buffer size ({}) was inferior to the needed size ({})."
                        .format(reserved_string_buffer_size, required_string_buffer_size.value)
                    )
3328
                self.__name_inner_eval = \
3329
                    [string_buffers[i].value.decode('utf-8') for i in range(self.__num_inner_eval)]
3330
                self.__higher_better_inner_eval = \
3331
                    [name.startswith(('auc', 'ndcg@', 'map@', 'average_precision')) for name in self.__name_inner_eval]
3332

wxchan's avatar
wxchan committed
3333
    def attr(self, key):
3334
        """Get attribute string from the Booster.
wxchan's avatar
wxchan committed
3335
3336
3337

        Parameters
        ----------
3338
3339
        key : string
            The name of the attribute.
wxchan's avatar
wxchan committed
3340
3341
3342

        Returns
        -------
3343
3344
        value : string or None
            The attribute value.
Nikita Titov's avatar
Nikita Titov committed
3345
            Returns None if attribute does not exist.
wxchan's avatar
wxchan committed
3346
        """
3347
        return self.__attr.get(key, None)
wxchan's avatar
wxchan committed
3348
3349

    def set_attr(self, **kwargs):
3350
        """Set attributes to the Booster.
wxchan's avatar
wxchan committed
3351
3352
3353
3354

        Parameters
        ----------
        **kwargs
3355
3356
            The attributes to set.
            Setting a value to None deletes an attribute.
Nikita Titov's avatar
Nikita Titov committed
3357
3358
3359
3360

        Returns
        -------
        self : Booster
3361
            Booster with set attributes.
wxchan's avatar
wxchan committed
3362
3363
3364
        """
        for key, value in kwargs.items():
            if value is not None:
3365
                if not isinstance(value, str):
Nikita Titov's avatar
Nikita Titov committed
3366
                    raise ValueError("Only string values are accepted")
wxchan's avatar
wxchan committed
3367
3368
3369
                self.__attr[key] = value
            else:
                self.__attr.pop(key, None)
Nikita Titov's avatar
Nikita Titov committed
3370
        return self