basic.py 85.7 KB
Newer Older
wxchan's avatar
wxchan committed
1
# coding: utf-8
2
3
# pylint: disable = invalid-name, C0111, C0301
# pylint: disable = R0912, R0913, R0914, W0105, W0201, W0212
wxchan's avatar
wxchan committed
4
5
6
"""Wrapper c_api of LightGBM"""
from __future__ import absolute_import

7
import copy
wxchan's avatar
wxchan committed
8
import ctypes
9
import os
wxchan's avatar
wxchan committed
10
import warnings
wxchan's avatar
wxchan committed
11
from tempfile import NamedTemporaryFile
wxchan's avatar
wxchan committed
12
13
14
15

import numpy as np
import scipy.sparse

16
17
18
19
from .compat import (DataFrame, LGBMDeprecationWarning, Series,
                     decode_string, integer_types,
                     json, json_default_with_numpy,
                     numeric_types, range_, string_type)
wxchan's avatar
wxchan committed
20
21
from .libpath import find_lib_path

wxchan's avatar
wxchan committed
22

wxchan's avatar
wxchan committed
23
24
25
26
def _load_lib():
    """Load LightGBM Library."""
    lib_path = find_lib_path()
    if len(lib_path) == 0:
27
        return None
wxchan's avatar
wxchan committed
28
29
30
31
    lib = ctypes.cdll.LoadLibrary(lib_path[0])
    lib.LGBM_GetLastError.restype = ctypes.c_char_p
    return lib

wxchan's avatar
wxchan committed
32

wxchan's avatar
wxchan committed
33
34
_LIB = _load_lib()

wxchan's avatar
wxchan committed
35

wxchan's avatar
wxchan committed
36
37
38
39
class LightGBMError(Exception):
    """Error throwed by LightGBM"""
    pass

wxchan's avatar
wxchan committed
40

wxchan's avatar
wxchan committed
41
42
43
44
45
46
47
48
def _safe_call(ret):
    """Check the return value of C API call
    Parameters
    ----------
    ret : int
        return value from API calls
    """
    if ret != 0:
49
        raise LightGBMError(decode_string(_LIB.LGBM_GetLastError()))
wxchan's avatar
wxchan committed
50

wxchan's avatar
wxchan committed
51

wxchan's avatar
wxchan committed
52
53
54
55
56
def is_numeric(obj):
    """Check is a number or not, include numpy number etc."""
    try:
        float(obj)
        return True
wxchan's avatar
wxchan committed
57
58
59
    except (TypeError, ValueError):
        # TypeError: obj is not a string or a number
        # ValueError: invalid literal
wxchan's avatar
wxchan committed
60
61
        return False

wxchan's avatar
wxchan committed
62

wxchan's avatar
wxchan committed
63
def is_numpy_1d_array(data):
Guolin Ke's avatar
Guolin Ke committed
64
    """Check is 1d numpy array"""
65
    return isinstance(data, np.ndarray) and len(data.shape) == 1
wxchan's avatar
wxchan committed
66

wxchan's avatar
wxchan committed
67

wxchan's avatar
wxchan committed
68
def is_1d_list(data):
Guolin Ke's avatar
Guolin Ke committed
69
    """Check is 1d list"""
70
    return isinstance(data, list) and \
71
        (not data or is_numeric(data[0]))
wxchan's avatar
wxchan committed
72

wxchan's avatar
wxchan committed
73

74
def list_to_1d_numpy(data, dtype=np.float32, name='list'):
Guolin Ke's avatar
Guolin Ke committed
75
    """convert to 1d numpy array"""
wxchan's avatar
wxchan committed
76
77
78
79
80
81
82
    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)
83
84
    elif isinstance(data, Series):
        return data.values.astype(dtype)
wxchan's avatar
wxchan committed
85
    else:
86
        raise TypeError("Wrong type({}) for {}, should be list or numpy array".format(type(data).__name__, name))
wxchan's avatar
wxchan committed
87

wxchan's avatar
wxchan committed
88

wxchan's avatar
wxchan committed
89
90
91
92
def cfloat32_array_to_numpy(cptr, length):
    """Convert a ctypes float pointer array to a numpy array.
    """
    if isinstance(cptr, ctypes.POINTER(ctypes.c_float)):
93
        return np.fromiter(cptr, dtype=np.float32, count=length)
wxchan's avatar
wxchan committed
94
    else:
95
        raise RuntimeError('Expected float pointer')
wxchan's avatar
wxchan committed
96

Guolin Ke's avatar
Guolin Ke committed
97

Guolin Ke's avatar
Guolin Ke committed
98
99
100
101
102
103
104
105
def cfloat64_array_to_numpy(cptr, length):
    """Convert a ctypes double pointer array to a numpy array.
    """
    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
106

wxchan's avatar
wxchan committed
107
108
109
110
def cint32_array_to_numpy(cptr, length):
    """Convert a ctypes float pointer array to a numpy array.
    """
    if isinstance(cptr, ctypes.POINTER(ctypes.c_int32)):
111
        return np.fromiter(cptr, dtype=np.int32, count=length)
wxchan's avatar
wxchan committed
112
    else:
113
        raise RuntimeError('Expected int pointer')
wxchan's avatar
wxchan committed
114

wxchan's avatar
wxchan committed
115

wxchan's avatar
wxchan committed
116
117
118
119
def c_str(string):
    """Convert a python string to cstring."""
    return ctypes.c_char_p(string.encode('utf-8'))

wxchan's avatar
wxchan committed
120

wxchan's avatar
wxchan committed
121
122
123
124
def c_array(ctype, values):
    """Convert a python array to c array."""
    return (ctype * len(values))(*values)

wxchan's avatar
wxchan committed
125

wxchan's avatar
wxchan committed
126
def param_dict_to_str(data):
127
    if data is None or not data:
wxchan's avatar
wxchan committed
128
129
130
        return ""
    pairs = []
    for key, val in data.items():
131
        if isinstance(val, (list, tuple, set)) or is_numpy_1d_array(val):
wxchan's avatar
wxchan committed
132
            pairs.append(str(key) + '=' + ','.join(map(str, val)))
wxchan's avatar
wxchan committed
133
        elif isinstance(val, string_type) or isinstance(val, numeric_types) or is_numeric(val):
wxchan's avatar
wxchan committed
134
            pairs.append(str(key) + '=' + str(val))
135
        elif val is not None:
136
            raise TypeError('Unknown type of parameter:%s, got:%s'
wxchan's avatar
wxchan committed
137
138
                            % (key, type(val).__name__))
    return ' '.join(pairs)
139

wxchan's avatar
wxchan committed
140

141
class _temp_file(object):
142
143
144
145
    def __enter__(self):
        with NamedTemporaryFile(prefix="lightgbm_tmp_", delete=True) as f:
            self.name = f.name
        return self
wxchan's avatar
wxchan committed
146

147
148
149
    def __exit__(self, exc_type, exc_val, exc_tb):
        if os.path.isfile(self.name):
            os.remove(self.name)
wxchan's avatar
wxchan committed
150

151
152
153
154
    def readlines(self):
        with open(self.name, "r+") as f:
            ret = f.readlines()
        return ret
wxchan's avatar
wxchan committed
155

156
157
    def writelines(self, lines):
        with open(self.name, "w+") as f:
158
            f.writelines(lines)
159

wxchan's avatar
wxchan committed
160

wxchan's avatar
wxchan committed
161
162
163
164
165
"""marco definition of data type in c_api of LightGBM"""
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
166

wxchan's avatar
wxchan committed
167
168
169
"""Matric is row major in python"""
C_API_IS_ROW_MAJOR = 1

Guolin Ke's avatar
Guolin Ke committed
170
"""marco definition of prediction type in c_api of LightGBM"""
wxchan's avatar
wxchan committed
171
172
173
C_API_PREDICT_NORMAL = 0
C_API_PREDICT_RAW_SCORE = 1
C_API_PREDICT_LEAF_INDEX = 2
174
C_API_PREDICT_CONTRIB = 3
wxchan's avatar
wxchan committed
175

Guolin Ke's avatar
Guolin Ke committed
176
"""data type of data field"""
wxchan's avatar
wxchan committed
177
178
FIELD_TYPE_MAPPER = {"label": C_API_DTYPE_FLOAT32,
                     "weight": C_API_DTYPE_FLOAT32,
Guolin Ke's avatar
Guolin Ke committed
179
                     "init_score": C_API_DTYPE_FLOAT64,
wxchan's avatar
wxchan committed
180
181
                     "group": C_API_DTYPE_INT32}

wxchan's avatar
wxchan committed
182

183
184
185
186
def convert_from_sliced_object(data):
    """fix the memory of multi-dimensional sliced object"""
    if data.base is not None and isinstance(data, np.ndarray) and isinstance(data.base, np.ndarray):
        if not data.flags.c_contiguous:
Nikita Titov's avatar
Nikita Titov committed
187
            warnings.warn("Usage of np.ndarray subset (sliced data) is not recommended due to it will double the peak memory cost in LightGBM.")
188
189
190
191
            return np.copy(data)
    return data


wxchan's avatar
wxchan committed
192
def c_float_array(data):
Guolin Ke's avatar
Guolin Ke committed
193
    """get pointer of float numpy array / list"""
wxchan's avatar
wxchan committed
194
195
196
    if is_1d_list(data):
        data = np.array(data, copy=False)
    if is_numpy_1d_array(data):
197
198
        data = convert_from_sliced_object(data)
        assert data.flags.c_contiguous
wxchan's avatar
wxchan committed
199
200
201
202
203
204
205
        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:
206
            raise TypeError("Expected np.float32 or np.float64, met type({})"
wxchan's avatar
wxchan committed
207
208
                            .format(data.dtype))
    else:
209
        raise TypeError("Unknown type({})".format(type(data).__name__))
210
    return (ptr_data, type_data, data)  # return `data` to avoid the temporary copy is freed
wxchan's avatar
wxchan committed
211

wxchan's avatar
wxchan committed
212

wxchan's avatar
wxchan committed
213
def c_int_array(data):
Guolin Ke's avatar
Guolin Ke committed
214
    """get pointer of int numpy array / list"""
wxchan's avatar
wxchan committed
215
216
217
    if is_1d_list(data):
        data = np.array(data, copy=False)
    if is_numpy_1d_array(data):
218
219
        data = convert_from_sliced_object(data)
        assert data.flags.c_contiguous
wxchan's avatar
wxchan committed
220
221
222
223
224
225
226
        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:
227
            raise TypeError("Expected np.int32 or np.int64, met type({})"
wxchan's avatar
wxchan committed
228
229
                            .format(data.dtype))
    else:
230
        raise TypeError("Unknown type({})".format(type(data).__name__))
231
    return (ptr_data, type_data, data)  # return `data` to avoid the temporary copy is freed
wxchan's avatar
wxchan committed
232

wxchan's avatar
wxchan committed
233

234
235
236
237
238
239
PANDAS_DTYPE_MAPPER = {'int8': 'int', 'int16': 'int', 'int32': 'int',
                       'int64': 'int', 'uint8': 'int', 'uint16': 'int',
                       'uint32': 'int', 'uint64': 'int', 'float16': 'float',
                       'float32': 'float', 'float64': 'float', 'bool': 'int'}


240
def _data_from_pandas(data, feature_name, categorical_feature, pandas_categorical):
241
    if isinstance(data, DataFrame):
242
243
        if len(data.shape) != 2 or data.shape[0] < 1:
            raise ValueError('Input data must be 2 dimensional and non empty.')
244
245
246
247
248
249
250
251
252
253
254
255
256
        if feature_name == 'auto' or feature_name is None:
            data = data.rename(columns=str)
        cat_cols = data.select_dtypes(include=['category']).columns
        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.')
            for col, category in zip(cat_cols, pandas_categorical):
                if list(data[col].cat.categories) != list(category):
                    data[col] = data[col].cat.set_categories(category)
        if len(cat_cols):  # cat_cols is pandas Index object
            data = data.copy()  # not alter origin DataFrame
257
            data[cat_cols] = data[cat_cols].apply(lambda x: x.cat.codes).replace({-1: np.nan})
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
        if categorical_feature is not None:
            if feature_name is None:
                feature_name = list(data.columns)
            if categorical_feature == 'auto':
                categorical_feature = list(cat_cols)
            else:
                categorical_feature = list(categorical_feature) + list(cat_cols)
        if feature_name == 'auto':
            feature_name = list(data.columns)
        data_dtypes = data.dtypes
        if not all(dtype.name in PANDAS_DTYPE_MAPPER for dtype in data_dtypes):
            bad_fields = [data.columns[i] for i, dtype in
                          enumerate(data_dtypes) if dtype.name not in PANDAS_DTYPE_MAPPER]

            msg = """DataFrame.dtypes for data must be int, float or bool. Did not expect the data types in fields """
            raise ValueError(msg + ', '.join(bad_fields))
274
        data = data.values.astype('float')
275
276
277
278
279
280
    else:
        if feature_name == 'auto':
            feature_name = None
        if categorical_feature == 'auto':
            categorical_feature = None
    return data, feature_name, categorical_feature, pandas_categorical
281
282
283
284
285
286
287
288
289
290
291
292
293


def _label_from_pandas(label):
    if isinstance(label, DataFrame):
        if len(label.columns) > 1:
            raise ValueError('DataFrame for label cannot have multiple columns')
        label_dtypes = label.dtypes
        if not all(dtype.name in PANDAS_DTYPE_MAPPER for dtype in label_dtypes):
            raise ValueError('DataFrame.dtypes for label must be int, float or bool')
        label = label.values.astype('float')
    return label


294
295
def _save_pandas_categorical(file_name, pandas_categorical):
    with open(file_name, 'a') as f:
Guolin Ke's avatar
Guolin Ke committed
296
        f.write('\npandas_categorical:' + json.dumps(pandas_categorical, default=json_default_with_numpy) + '\n')
297
298
299
300


def _load_pandas_categorical(file_name):
    with open(file_name, 'r') as f:
Guolin Ke's avatar
Guolin Ke committed
301
302
303
304
        lines = f.readlines()
        last_line = lines[-1]
        if last_line.strip() == "":
            last_line = lines[-2]
305
306
307
308
309
        if last_line.startswith('pandas_categorical:'):
            return json.loads(last_line[len('pandas_categorical:'):])
    return None


Guolin Ke's avatar
Guolin Ke committed
310
311
class _InnerPredictor(object):
    """
312
313
    A _InnerPredictor of LightGBM.
    Only used for prediction, usually used for continued-train
Guolin Ke's avatar
Guolin Ke committed
314
    Note: Can convert from Booster, but cannot convert to Booster
wxchan's avatar
wxchan committed
315
    """
316
    def __init__(self, model_file=None, booster_handle=None, pred_parameter=None):
Guolin Ke's avatar
Guolin Ke committed
317
        """Initialize the _InnerPredictor. Not expose to user
wxchan's avatar
wxchan committed
318
319
320
321
322

        Parameters
        ----------
        model_file : string
            Path to the model file.
Guolin Ke's avatar
Guolin Ke committed
323
324
        booster_handle : Handle of Booster
            use handle to init
325
326
        pred_parameter: dict
            Other parameters for the prediciton
wxchan's avatar
wxchan committed
327
328
329
330
331
        """
        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
332
            out_num_iterations = ctypes.c_int(0)
wxchan's avatar
wxchan committed
333
334
335
336
            _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
337
            out_num_class = ctypes.c_int(0)
wxchan's avatar
wxchan committed
338
339
340
341
            _safe_call(_LIB.LGBM_BoosterGetNumClasses(
                self.handle,
                ctypes.byref(out_num_class)))
            self.num_class = out_num_class.value
342
            self.num_total_iteration = out_num_iterations.value
343
            self.pandas_categorical = _load_pandas_categorical(model_file)
wxchan's avatar
wxchan committed
344
        elif booster_handle is not None:
Guolin Ke's avatar
Guolin Ke committed
345
            self.__is_manage_handle = False
wxchan's avatar
wxchan committed
346
            self.handle = booster_handle
Guolin Ke's avatar
Guolin Ke committed
347
            out_num_class = ctypes.c_int(0)
wxchan's avatar
wxchan committed
348
349
350
351
            _safe_call(_LIB.LGBM_BoosterGetNumClasses(
                self.handle,
                ctypes.byref(out_num_class)))
            self.num_class = out_num_class.value
Guolin Ke's avatar
Guolin Ke committed
352
            out_num_iterations = ctypes.c_int(0)
wxchan's avatar
wxchan committed
353
354
355
            _safe_call(_LIB.LGBM_BoosterGetCurrentIteration(
                self.handle,
                ctypes.byref(out_num_iterations)))
356
            self.num_total_iteration = out_num_iterations.value
357
            self.pandas_categorical = None
wxchan's avatar
wxchan committed
358
        else:
Guolin Ke's avatar
Guolin Ke committed
359
            raise TypeError('Need Model file or Booster handle to create a predictor')
wxchan's avatar
wxchan committed
360

361
362
        pred_parameter = {} if pred_parameter is None else pred_parameter
        self.pred_parameter = param_dict_to_str(pred_parameter)
cbecker's avatar
cbecker committed
363

wxchan's avatar
wxchan committed
364
    def __del__(self):
365
366
367
368
369
        try:
            if self.__is_manage_handle:
                _safe_call(_LIB.LGBM_BoosterFree(self.handle))
        except AttributeError:
            pass
wxchan's avatar
wxchan committed
370

371
372
373
374
375
    def __getstate__(self):
        this = self.__dict__.copy()
        this.pop('handle', None)
        return this

wxchan's avatar
wxchan committed
376
    def predict(self, data, num_iteration=-1,
377
                raw_score=False, pred_leaf=False, pred_contrib=False, data_has_header=False,
wxchan's avatar
wxchan committed
378
379
380
381
382
383
384
385
                is_reshape=True):
        """
        Predict logic

        Parameters
        ----------
        data : string/numpy array/scipy.sparse
            Data source for prediction
386
            When data type is string, it represents the path of txt file
wxchan's avatar
wxchan committed
387
        num_iteration : int
388
            Used iteration for prediction
wxchan's avatar
wxchan committed
389
390
391
392
        raw_score : bool
            True for predict raw score
        pred_leaf : bool
            True for predict leaf index
393
394
        pred_contrib : bool
            True for predict feature contributions
wxchan's avatar
wxchan committed
395
        data_has_header : bool
Guolin Ke's avatar
Guolin Ke committed
396
            Used for txt data, True if txt data has header
wxchan's avatar
wxchan committed
397
        is_reshape : bool
398
            Reshape to (nrow, ncol) if true
wxchan's avatar
wxchan committed
399
400
401
402
403

        Returns
        -------
        Prediction result
        """
wxchan's avatar
wxchan committed
404
        if isinstance(data, Dataset):
405
            raise TypeError("Cannot use Dataset instance for prediction, please use raw data instead")
406
        data = _data_from_pandas(data, None, None, self.pandas_categorical)[0]
wxchan's avatar
wxchan committed
407
408
409
410
411
        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
412
413
        if pred_contrib:
            predict_type = C_API_PREDICT_CONTRIB
wxchan's avatar
wxchan committed
414
        int_data_has_header = 1 if data_has_header else 0
415
416
        if num_iteration > self.num_total_iteration:
            num_iteration = self.num_total_iteration
cbecker's avatar
cbecker committed
417

wxchan's avatar
wxchan committed
418
        if isinstance(data, string_type):
419
            with _temp_file() as f:
wxchan's avatar
wxchan committed
420
421
422
                _safe_call(_LIB.LGBM_BoosterPredictForFile(
                    self.handle,
                    c_str(data),
Guolin Ke's avatar
Guolin Ke committed
423
424
425
                    ctypes.c_int(int_data_has_header),
                    ctypes.c_int(predict_type),
                    ctypes.c_int(num_iteration),
426
                    c_str(self.pred_parameter),
wxchan's avatar
wxchan committed
427
428
                    c_str(f.name)))
                lines = f.readlines()
429
430
                nrow = len(lines)
                preds = [float(token) for line in lines for token in line.split('\t')]
Guolin Ke's avatar
Guolin Ke committed
431
                preds = np.array(preds, dtype=np.float64, copy=False)
wxchan's avatar
wxchan committed
432
433
434
        elif isinstance(data, scipy.sparse.csr_matrix):
            preds, nrow = self.__pred_for_csr(data, num_iteration,
                                              predict_type)
Guolin Ke's avatar
Guolin Ke committed
435
436
437
        elif isinstance(data, scipy.sparse.csc_matrix):
            preds, nrow = self.__pred_for_csc(data, num_iteration,
                                              predict_type)
wxchan's avatar
wxchan committed
438
439
440
        elif isinstance(data, np.ndarray):
            preds, nrow = self.__pred_for_np2d(data, num_iteration,
                                               predict_type)
441
442
443
        elif isinstance(data, list):
            try:
                data = np.array(data)
444
            except BaseException:
445
446
447
                raise ValueError('Cannot convert data list to numpy array.')
            preds, nrow = self.__pred_for_np2d(data, num_iteration,
                                               predict_type)
wxchan's avatar
wxchan committed
448
449
        else:
            try:
450
                warnings.warn('Converting data to scipy sparse matrix.')
wxchan's avatar
wxchan committed
451
                csr = scipy.sparse.csr_matrix(data)
452
            except BaseException:
453
                raise TypeError('Cannot predict data for type {}'.format(type(data).__name__))
454
455
            preds, nrow = self.__pred_for_csr(csr, num_iteration,
                                              predict_type)
wxchan's avatar
wxchan committed
456
457
        if pred_leaf:
            preds = preds.astype(np.int32)
458
        if is_reshape and preds.size != nrow:
wxchan's avatar
wxchan committed
459
            if preds.size % nrow == 0:
460
                preds = preds.reshape(nrow, -1)
wxchan's avatar
wxchan committed
461
            else:
462
                raise ValueError('Length of predict result (%d) cannot be divide nrow (%d)'
wxchan's avatar
wxchan committed
463
464
465
466
                                 % (preds.size, nrow))
        return preds

    def __get_num_preds(self, num_iteration, nrow, predict_type):
Guolin Ke's avatar
Guolin Ke committed
467
468
469
        """
        Get size of prediction result
        """
Guolin Ke's avatar
Guolin Ke committed
470
471
472
        n_preds = ctypes.c_int64(0)
        _safe_call(_LIB.LGBM_BoosterCalcNumPredict(
            self.handle,
Guolin Ke's avatar
Guolin Ke committed
473
474
475
            ctypes.c_int(nrow),
            ctypes.c_int(predict_type),
            ctypes.c_int(num_iteration),
Guolin Ke's avatar
Guolin Ke committed
476
477
            ctypes.byref(n_preds)))
        return n_preds.value
wxchan's avatar
wxchan committed
478
479
480
481
482
483

    def __pred_for_np2d(self, mat, num_iteration, predict_type):
        """
        Predict for a 2-D numpy matrix.
        """
        if len(mat.shape) != 2:
484
            raise ValueError('Input numpy.ndarray or list must be 2 dimensional')
wxchan's avatar
wxchan committed
485
486
487
488
489
490

        if mat.dtype == np.float32 or mat.dtype == np.float64:
            data = np.array(mat.reshape(mat.size), dtype=mat.dtype, copy=False)
        else:
            """change non-float data to float data, need to copy"""
            data = np.array(mat.reshape(mat.size), dtype=np.float32)
491
        ptr_data, type_ptr_data, _ = c_float_array(data)
wxchan's avatar
wxchan committed
492
493
        n_preds = self.__get_num_preds(num_iteration, mat.shape[0],
                                       predict_type)
Guolin Ke's avatar
Guolin Ke committed
494
        preds = np.zeros(n_preds, dtype=np.float64)
wxchan's avatar
wxchan committed
495
496
497
498
        out_num_preds = ctypes.c_int64(0)
        _safe_call(_LIB.LGBM_BoosterPredictForMat(
            self.handle,
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
499
500
501
502
503
504
            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),
            ctypes.c_int(num_iteration),
505
            c_str(self.pred_parameter),
wxchan's avatar
wxchan committed
506
            ctypes.byref(out_num_preds),
wxchan's avatar
wxchan committed
507
            preds.ctypes.data_as(ctypes.POINTER(ctypes.c_double))))
wxchan's avatar
wxchan committed
508
        if n_preds != out_num_preds.value:
509
            raise ValueError("Wrong length for predict results")
wxchan's avatar
wxchan committed
510
511
512
513
514
515
516
517
        return preds, mat.shape[0]

    def __pred_for_csr(self, csr, num_iteration, predict_type):
        """
        Predict for a csr data
        """
        nrow = len(csr.indptr) - 1
        n_preds = self.__get_num_preds(num_iteration, nrow, predict_type)
Guolin Ke's avatar
Guolin Ke committed
518
        preds = np.zeros(n_preds, dtype=np.float64)
wxchan's avatar
wxchan committed
519
520
        out_num_preds = ctypes.c_int64(0)

521
522
        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
523
524
525
526

        _safe_call(_LIB.LGBM_BoosterPredictForCSR(
            self.handle,
            ptr_indptr,
Guolin Ke's avatar
Guolin Ke committed
527
            ctypes.c_int32(type_ptr_indptr),
wxchan's avatar
wxchan committed
528
529
            csr.indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
530
531
532
533
534
535
            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),
            ctypes.c_int(num_iteration),
536
            c_str(self.pred_parameter),
wxchan's avatar
wxchan committed
537
            ctypes.byref(out_num_preds),
wxchan's avatar
wxchan committed
538
            preds.ctypes.data_as(ctypes.POINTER(ctypes.c_double))))
Guolin Ke's avatar
Guolin Ke committed
539
540
541
542
543
544
545
546
547
548
549
550
551
        if n_preds != out_num_preds.value:
            raise ValueError("Wrong length for predict results")
        return preds, nrow

    def __pred_for_csc(self, csc, num_iteration, predict_type):
        """
        Predict for a csc data
        """
        nrow = csc.shape[0]
        n_preds = self.__get_num_preds(num_iteration, nrow, predict_type)
        preds = np.zeros(n_preds, dtype=np.float64)
        out_num_preds = ctypes.c_int64(0)

552
553
        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
554
555
556
557

        _safe_call(_LIB.LGBM_BoosterPredictForCSC(
            self.handle,
            ptr_indptr,
Guolin Ke's avatar
Guolin Ke committed
558
            ctypes.c_int32(type_ptr_indptr),
Guolin Ke's avatar
Guolin Ke committed
559
560
            csc.indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
561
562
563
564
565
566
            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),
            ctypes.c_int(num_iteration),
567
            c_str(self.pred_parameter),
Guolin Ke's avatar
Guolin Ke committed
568
            ctypes.byref(out_num_preds),
wxchan's avatar
wxchan committed
569
            preds.ctypes.data_as(ctypes.POINTER(ctypes.c_double))))
wxchan's avatar
wxchan committed
570
        if n_preds != out_num_preds.value:
571
            raise ValueError("Wrong length for predict results")
wxchan's avatar
wxchan committed
572
573
        return preds, nrow

wxchan's avatar
wxchan committed
574

wxchan's avatar
wxchan committed
575
576
class Dataset(object):
    """Dataset in LightGBM."""
577
    def __init__(self, data, label=None, reference=None,
578
                 weight=None, group=None, init_score=None, silent=False,
579
                 feature_name='auto', categorical_feature='auto', params=None,
wxchan's avatar
wxchan committed
580
                 free_raw_data=True):
581
582
        """Constract Dataset.

wxchan's avatar
wxchan committed
583
584
        Parameters
        ----------
585
        data : string, numpy array, scipy.sparse or list of numpy arrays
wxchan's avatar
wxchan committed
586
            Data source of Dataset.
587
            If string, it represents the path to txt file.
588
        label : list, numpy 1-D array or None, optional (default=None)
589
590
591
592
            Label of the data.
        reference : Dataset or None, optional (default=None)
            If this is Dataset for validation, training data should be used as reference.
        weight : list, numpy 1-D array or None, optional (default=None)
wxchan's avatar
wxchan committed
593
            Weight for each instance.
594
595
        group : list, numpy 1-D array or None, optional (default=None)
            Group/query size for Dataset.
596
597
        init_score : list, numpy 1-D array or None, optional (default=None)
            Init score for Dataset.
598
599
600
601
602
603
604
605
606
607
        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).
            If 'auto' and data is pandas DataFrame, pandas categorical columns are used.
608
609
            All values in categorical features should be less than int32 max value (2147483647).
            All negative values in categorical features will be treated as missing values.
Nikita Titov's avatar
Nikita Titov committed
610
        params : dict or None, optional (default=None)
611
            Other parameters.
Nikita Titov's avatar
Nikita Titov committed
612
        free_raw_data : bool, optional (default=True)
613
            If True, raw data is freed after constructing inner Dataset.
wxchan's avatar
wxchan committed
614
        """
wxchan's avatar
wxchan committed
615
616
617
618
619
620
        self.handle = None
        self.data = data
        self.label = label
        self.reference = reference
        self.weight = weight
        self.group = group
621
        self.init_score = init_score
wxchan's avatar
wxchan committed
622
623
        self.silent = silent
        self.feature_name = feature_name
624
        self.categorical_feature = categorical_feature
625
        self.params = copy.deepcopy(params)
wxchan's avatar
wxchan committed
626
627
628
        self.free_raw_data = free_raw_data
        self.used_indices = None
        self._predictor = None
629
        self.pandas_categorical = None
630
        self.params_back_up = None
wxchan's avatar
wxchan committed
631
632

    def __del__(self):
633
634
635
636
        try:
            self._free_handle()
        except AttributeError:
            pass
637
638

    def _free_handle(self):
639
        if self.handle is not None:
640
            _safe_call(_LIB.LGBM_DatasetFree(self.handle))
641
            self.handle = None
Nikita Titov's avatar
Nikita Titov committed
642
        return self
wxchan's avatar
wxchan committed
643

644
    def _lazy_init(self, data, label=None, reference=None,
645
                   weight=None, group=None, init_score=None, predictor=None,
wxchan's avatar
wxchan committed
646
                   silent=False, feature_name='auto',
647
                   categorical_feature='auto', params=None):
wxchan's avatar
wxchan committed
648
649
        if data is None:
            self.handle = None
Nikita Titov's avatar
Nikita Titov committed
650
            return self
Guolin Ke's avatar
Guolin Ke committed
651
652
653
        if reference is not None:
            self.pandas_categorical = reference.pandas_categorical
            categorical_feature = reference.categorical_feature
654
        data, feature_name, categorical_feature, self.pandas_categorical = _data_from_pandas(data, feature_name, categorical_feature, self.pandas_categorical)
wxchan's avatar
wxchan committed
655
656
        label = _label_from_pandas(label)
        self.data_has_header = False
657
        # process for args
wxchan's avatar
wxchan committed
658
        params = {} if params is None else params
659
660
661
662
663
        args_names = getattr(self.__class__, '_lazy_init').__code__.co_varnames[:getattr(self.__class__, '_lazy_init').__code__.co_argcount]
        for key, _ in params.items():
            if key in args_names:
                warnings.warn('{0} keyword has been found in `params` and will be ignored. '
                              'Please use {0} argument of the Dataset constructor to pass this parameter.'.format(key))
wxchan's avatar
wxchan committed
664
665
666
        self.predictor = predictor
        if silent:
            params["verbose"] = 0
667
        # get categorical features
668
669
670
671
672
673
674
675
676
677
678
679
680
        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:
                if isinstance(name, string_type) and name in feature_dict:
                    categorical_indices.add(feature_dict[name])
                elif isinstance(name, integer_types):
                    categorical_indices.add(name)
                else:
                    raise TypeError("Wrong type({}) or unknown name({}) in categorical_feature"
                                    .format(type(name).__name__, name))
681
            if categorical_indices:
682
                if "categorical_feature" in params or "categorical_column" in params:
683
                    warnings.warn('categorical_feature in param dict is overridden.')
684
685
                    params.pop("categorical_feature", None)
                    params.pop("categorical_column", None)
686
                params['categorical_column'] = sorted(categorical_indices)
687

wxchan's avatar
wxchan committed
688
        params_str = param_dict_to_str(params)
689
        # process for reference dataset
wxchan's avatar
wxchan committed
690
        ref_dataset = None
wxchan's avatar
wxchan committed
691
        if isinstance(reference, Dataset):
692
            ref_dataset = reference.construct().handle
wxchan's avatar
wxchan committed
693
694
        elif reference is not None:
            raise TypeError('Reference dataset should be None or dataset instance')
695
        # start construct data
wxchan's avatar
wxchan committed
696
        if isinstance(data, string_type):
697
            # check data has header or not
Guolin Ke's avatar
Guolin Ke committed
698
            if str(params.get("has_header", "")).lower() == "true" \
wxchan's avatar
wxchan committed
699
                    or str(params.get("header", "")).lower() == "true":
700
                self.data_has_header = True
wxchan's avatar
wxchan committed
701
702
703
704
705
706
707
708
            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
709
710
        elif isinstance(data, scipy.sparse.csc_matrix):
            self.__init_from_csc(data, params_str, ref_dataset)
wxchan's avatar
wxchan committed
711
712
        elif isinstance(data, np.ndarray):
            self.__init_from_np2d(data, params_str, ref_dataset)
713
714
        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)
wxchan's avatar
wxchan committed
715
716
717
718
        else:
            try:
                csr = scipy.sparse.csr_matrix(data)
                self.__init_from_csr(csr, params_str, ref_dataset)
719
            except BaseException:
wxchan's avatar
wxchan committed
720
                raise TypeError('Cannot initialize Dataset from {}'.format(type(data).__name__))
wxchan's avatar
wxchan committed
721
722
723
        if label is not None:
            self.set_label(label)
        if self.get_label() is None:
724
            raise ValueError("Label should not be None")
wxchan's avatar
wxchan committed
725
726
727
728
729
        if weight is not None:
            self.set_weight(weight)
        if group is not None:
            self.set_group(group)
        # load init score
730
731
732
        if init_score is not None:
            self.set_init_score(init_score)
            if self.predictor is not None:
733
                warnings.warn("The prediction of init_model will be overridden by init_score.")
734
        elif isinstance(self.predictor, _InnerPredictor):
wxchan's avatar
wxchan committed
735
736
737
738
739
740
            init_score = self.predictor.predict(data,
                                                raw_score=True,
                                                data_has_header=self.data_has_header,
                                                is_reshape=False)
            if self.predictor.num_class > 1:
                # need re group init score
wxchan's avatar
wxchan committed
741
                new_init_score = np.zeros(init_score.size, dtype=np.float32)
wxchan's avatar
wxchan committed
742
                num_data = self.num_data()
wxchan's avatar
wxchan committed
743
744
                for i in range_(num_data):
                    for j in range_(self.predictor.num_class):
wxchan's avatar
wxchan committed
745
746
747
                        new_init_score[j * num_data + i] = init_score[i * self.predictor.num_class + j]
                init_score = new_init_score
            self.set_init_score(init_score)
Guolin Ke's avatar
Guolin Ke committed
748
749
        elif self.predictor is not None:
            raise TypeError('wrong predictor type {}'.format(type(self.predictor).__name__))
Guolin Ke's avatar
Guolin Ke committed
750
        # set feature names
Nikita Titov's avatar
Nikita Titov committed
751
        return self.set_feature_name(feature_name)
wxchan's avatar
wxchan committed
752
753
754
755
756
757
758
759
760
761
762
763

    def __init_from_np2d(self, mat, params_str, ref_dataset):
        """
        Initialize data from a 2-D numpy matrix.
        """
        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)
        else:
764
            # change non-float data to float data, need to copy
wxchan's avatar
wxchan committed
765
766
            data = np.array(mat.reshape(mat.size), dtype=np.float32)

767
        ptr_data, type_ptr_data, _ = c_float_array(data)
wxchan's avatar
wxchan committed
768
769
        _safe_call(_LIB.LGBM_DatasetCreateFromMat(
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
770
771
772
773
            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
774
775
776
            c_str(params_str),
            ref_dataset,
            ctypes.byref(self.handle)))
Nikita Titov's avatar
Nikita Titov committed
777
        return self
wxchan's avatar
wxchan committed
778

779
780
781
782
783
784
785
786
787
788
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
814
815
816
817
818
819
820
821
822
823
824
825
    def __init_from_list_np2d(self, mats, params_str, ref_dataset):
        """
        Initialize data from list of 2-D numpy matrices.
        """
        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)
            else:
                # change non-float data to float data, need to copy
                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
826
        return self
827

wxchan's avatar
wxchan committed
828
829
830
831
832
    def __init_from_csr(self, csr, params_str, ref_dataset):
        """
        Initialize data from a CSR matrix.
        """
        if len(csr.indices) != len(csr.data):
833
            raise ValueError('Length mismatch: {} vs {}'.format(len(csr.indices), len(csr.data)))
wxchan's avatar
wxchan committed
834
835
        self.handle = ctypes.c_void_p()

836
837
        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
838
839
840

        _safe_call(_LIB.LGBM_DatasetCreateFromCSR(
            ptr_indptr,
Guolin Ke's avatar
Guolin Ke committed
841
            ctypes.c_int(type_ptr_indptr),
wxchan's avatar
wxchan committed
842
843
            csr.indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
844
845
846
847
            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
848
849
850
            c_str(params_str),
            ref_dataset,
            ctypes.byref(self.handle)))
Nikita Titov's avatar
Nikita Titov committed
851
        return self
wxchan's avatar
wxchan committed
852

Guolin Ke's avatar
Guolin Ke committed
853
854
855
856
857
858
859
860
    def __init_from_csc(self, csc, params_str, ref_dataset):
        """
        Initialize data from a csc matrix.
        """
        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()

861
862
        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
863
864
865

        _safe_call(_LIB.LGBM_DatasetCreateFromCSC(
            ptr_indptr,
Guolin Ke's avatar
Guolin Ke committed
866
            ctypes.c_int(type_ptr_indptr),
Guolin Ke's avatar
Guolin Ke committed
867
868
            csc.indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
869
870
871
872
            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
873
874
875
            c_str(params_str),
            ref_dataset,
            ctypes.byref(self.handle)))
Nikita Titov's avatar
Nikita Titov committed
876
        return self
Guolin Ke's avatar
Guolin Ke committed
877

wxchan's avatar
wxchan committed
878
    def construct(self):
879
880
881
882
883
        """Lazy init.

        Returns
        -------
        self : Dataset
Nikita Titov's avatar
Nikita Titov committed
884
            Constructed Dataset object.
885
        """
886
        if self.handle is None:
wxchan's avatar
wxchan committed
887
888
            if self.reference is not None:
                if self.used_indices is None:
889
                    # create valid
890
                    self._lazy_init(self.data, label=self.label, reference=self.reference,
891
                                    weight=self.weight, group=self.group, init_score=self.init_score, predictor=self._predictor,
892
                                    silent=self.silent, feature_name=self.feature_name, params=self.params)
wxchan's avatar
wxchan committed
893
                else:
894
                    # construct subset
wxchan's avatar
wxchan committed
895
                    used_indices = list_to_1d_numpy(self.used_indices, np.int32, name='used_indices')
896
                    assert used_indices.flags.c_contiguous
897
                    self.handle = ctypes.c_void_p()
wxchan's avatar
wxchan committed
898
899
                    params_str = param_dict_to_str(self.params)
                    _safe_call(_LIB.LGBM_DatasetGetSubset(
900
                        self.reference.construct().handle,
wxchan's avatar
wxchan committed
901
                        used_indices.ctypes.data_as(ctypes.POINTER(ctypes.c_int32)),
Guolin Ke's avatar
Guolin Ke committed
902
                        ctypes.c_int(used_indices.shape[0]),
wxchan's avatar
wxchan committed
903
904
905
906
907
                        c_str(params_str),
                        ctypes.byref(self.handle)))
                    if self.get_label() is None:
                        raise ValueError("Label should not be None.")
            else:
908
                # create train
909
                self._lazy_init(self.data, label=self.label,
910
911
                                weight=self.weight, group=self.group, init_score=self.init_score,
                                predictor=self._predictor, silent=self.silent, feature_name=self.feature_name,
912
                                categorical_feature=self.categorical_feature, params=self.params)
wxchan's avatar
wxchan committed
913
914
915
            if self.free_raw_data:
                self.data = None
        return self
wxchan's avatar
wxchan committed
916

wxchan's avatar
wxchan committed
917
    def create_valid(self, data, label=None, weight=None, group=None,
918
                     init_score=None, silent=False, params=None):
919
        """Create validation data align with current Dataset.
wxchan's avatar
wxchan committed
920
921
922

        Parameters
        ----------
923
        data : string, numpy array or scipy.sparse
wxchan's avatar
wxchan committed
924
            Data source of Dataset.
925
926
            If string, it represents the path to txt file.
        label : list or numpy 1-D array, optional (default=None)
wxchan's avatar
wxchan committed
927
            Label of the training data.
928
        weight : list, numpy 1-D array or None, optional (default=None)
wxchan's avatar
wxchan committed
929
            Weight for each instance.
930
931
        group : list, numpy 1-D array or None, optional (default=None)
            Group/query size for Dataset.
932
933
        init_score : list, numpy 1-D array or None, optional (default=None)
            Init score for Dataset.
934
935
        silent : bool, optional (default=False)
            Whether to print messages during construction.
Nikita Titov's avatar
Nikita Titov committed
936
        params : dict or None, optional (default=None)
937
938
939
940
            Other parameters.

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
941
942
        valid : Dataset
            Validation Dataset with reference to self.
wxchan's avatar
wxchan committed
943
        """
944
        ret = Dataset(data, label=label, reference=self,
945
946
                      weight=weight, group=group, init_score=init_score,
                      silent=silent, params=params, free_raw_data=self.free_raw_data)
wxchan's avatar
wxchan committed
947
        ret._predictor = self._predictor
948
        ret.pandas_categorical = self.pandas_categorical
wxchan's avatar
wxchan committed
949
        return ret
wxchan's avatar
wxchan committed
950

wxchan's avatar
wxchan committed
951
    def subset(self, used_indices, params=None):
952
        """Get subset of current Dataset.
wxchan's avatar
wxchan committed
953
954
955
956

        Parameters
        ----------
        used_indices : list of int
957
            Indices used to create the subset.
Nikita Titov's avatar
Nikita Titov committed
958
        params : dict or None, optional (default=None)
959
960
961
962
963
964
            Other parameters.

        Returns
        -------
        subset : Dataset
            Subset of the current Dataset.
wxchan's avatar
wxchan committed
965
        """
wxchan's avatar
wxchan committed
966
967
        if params is None:
            params = self.params
wxchan's avatar
wxchan committed
968
        ret = Dataset(None, reference=self, feature_name=self.feature_name,
969
                      categorical_feature=self.categorical_feature, params=params)
wxchan's avatar
wxchan committed
970
        ret._predictor = self._predictor
971
        ret.pandas_categorical = self.pandas_categorical
wxchan's avatar
wxchan committed
972
973
974
975
        ret.used_indices = used_indices
        return ret

    def save_binary(self, filename):
976
        """Save Dataset to binary file.
wxchan's avatar
wxchan committed
977
978
979
980
981

        Parameters
        ----------
        filename : string
            Name of the output file.
Nikita Titov's avatar
Nikita Titov committed
982
983
984
985
986

        Returns
        -------
        self : Dataset
            Returns self.
wxchan's avatar
wxchan committed
987
988
989
990
        """
        _safe_call(_LIB.LGBM_DatasetSaveBinary(
            self.construct().handle,
            c_str(filename)))
Nikita Titov's avatar
Nikita Titov committed
991
        return self
wxchan's avatar
wxchan committed
992
993
994
995

    def _update_params(self, params):
        if not self.params:
            self.params = params
wxchan's avatar
wxchan committed
996
        else:
997
            self.params_back_up = copy.deepcopy(self.params)
wxchan's avatar
wxchan committed
998
            self.params.update(params)
Nikita Titov's avatar
Nikita Titov committed
999
        return self
wxchan's avatar
wxchan committed
1000

1001
1002
1003
    def _reverse_update_params(self):
        self.params = copy.deepcopy(self.params_back_up)
        self.params_back_up = None
Nikita Titov's avatar
Nikita Titov committed
1004
        return self
1005

wxchan's avatar
wxchan committed
1006
    def set_field(self, field_name, data):
wxchan's avatar
wxchan committed
1007
        """Set property into the Dataset.
wxchan's avatar
wxchan committed
1008
1009
1010

        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
1011
        field_name : string
1012
            The field name of the information.
Nikita Titov's avatar
Nikita Titov committed
1013
        data : list, numpy array or None
1014
            The array of data to be set.
Nikita Titov's avatar
Nikita Titov committed
1015
1016
1017
1018
1019

        Returns
        -------
        self : Dataset
            Dataset with set property.
wxchan's avatar
wxchan committed
1020
        """
1021
1022
        if self.handle is None:
            raise Exception("Cannot set %s before construct dataset" % field_name)
wxchan's avatar
wxchan committed
1023
        if data is None:
1024
            # set to None
wxchan's avatar
wxchan committed
1025
1026
1027
1028
            _safe_call(_LIB.LGBM_DatasetSetField(
                self.handle,
                c_str(field_name),
                None,
Guolin Ke's avatar
Guolin Ke committed
1029
1030
                ctypes.c_int(0),
                ctypes.c_int(FIELD_TYPE_MAPPER[field_name])))
Nikita Titov's avatar
Nikita Titov committed
1031
            return self
Guolin Ke's avatar
Guolin Ke committed
1032
1033
1034
1035
1036
        dtype = np.float32
        if field_name == 'group':
            dtype = np.int32
        elif field_name == 'init_score':
            dtype = np.float64
1037
        data = list_to_1d_numpy(data, dtype, name=field_name)
1038
1039
        if data.dtype == np.float32 or data.dtype == np.float64:
            ptr_data, type_data, _ = c_float_array(data)
wxchan's avatar
wxchan committed
1040
        elif data.dtype == np.int32:
1041
            ptr_data, type_data, _ = c_int_array(data)
wxchan's avatar
wxchan committed
1042
        else:
Guolin Ke's avatar
Guolin Ke committed
1043
            raise TypeError("Excepted np.float32/64 or np.int32, meet type({})".format(data.dtype))
wxchan's avatar
wxchan committed
1044
        if type_data != FIELD_TYPE_MAPPER[field_name]:
1045
            raise TypeError("Input type error for set_field")
wxchan's avatar
wxchan committed
1046
1047
1048
1049
        _safe_call(_LIB.LGBM_DatasetSetField(
            self.handle,
            c_str(field_name),
            ptr_data,
Guolin Ke's avatar
Guolin Ke committed
1050
1051
            ctypes.c_int(len(data)),
            ctypes.c_int(type_data)))
Nikita Titov's avatar
Nikita Titov committed
1052
        return self
wxchan's avatar
wxchan committed
1053

wxchan's avatar
wxchan committed
1054
1055
    def get_field(self, field_name):
        """Get property from the Dataset.
wxchan's avatar
wxchan committed
1056
1057
1058

        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
1059
        field_name : string
1060
            The field name of the information.
wxchan's avatar
wxchan committed
1061
1062
1063

        Returns
        -------
1064
1065
        info : numpy array
            A numpy array with information from the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1066
        """
1067
        if self.handle is None:
1068
            raise Exception("Cannot get %s before construct Dataset" % field_name)
Guolin Ke's avatar
Guolin Ke committed
1069
1070
        tmp_out_len = ctypes.c_int()
        out_type = ctypes.c_int()
wxchan's avatar
wxchan committed
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
        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
1086
1087
        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)
1088
        else:
wxchan's avatar
wxchan committed
1089
            raise TypeError("Unknown type")
Guolin Ke's avatar
Guolin Ke committed
1090

1091
    def set_categorical_feature(self, categorical_feature):
1092
        """Set categorical features.
1093
1094
1095

        Parameters
        ----------
1096
1097
        categorical_feature : list of int or strings
            Names or indices of categorical features.
Nikita Titov's avatar
Nikita Titov committed
1098
1099
1100
1101
1102

        Returns
        -------
        self : Dataset
            Dataset with set categorical features.
1103
1104
        """
        if self.categorical_feature == categorical_feature:
Nikita Titov's avatar
Nikita Titov committed
1105
            return self
1106
        if self.data is not None:
1107
1108
            if self.categorical_feature is None:
                self.categorical_feature = categorical_feature
Nikita Titov's avatar
Nikita Titov committed
1109
                return self._free_handle()
1110
1111
            elif categorical_feature == 'auto':
                warnings.warn('Using categorical_feature in Dataset.')
Nikita Titov's avatar
Nikita Titov committed
1112
                return self
1113
            else:
1114
                warnings.warn('categorical_feature in Dataset is overridden. New categorical_feature is {}'.format(sorted(list(categorical_feature))))
1115
                self.categorical_feature = categorical_feature
Nikita Titov's avatar
Nikita Titov committed
1116
                return self._free_handle()
1117
1118
1119
        else:
            raise LightGBMError("Cannot set categorical feature after freed raw data, set free_raw_data=False when construct Dataset to avoid this.")

Guolin Ke's avatar
Guolin Ke committed
1120
1121
    def _set_predictor(self, predictor):
        """
1122
        Set predictor for continued training, not recommended for user to call this function.
Guolin Ke's avatar
Guolin Ke committed
1123
1124
1125
        Please set init_model in engine.train or engine.cv
        """
        if predictor is self._predictor:
Nikita Titov's avatar
Nikita Titov committed
1126
            return self
Guolin Ke's avatar
Guolin Ke committed
1127
1128
        if self.data is not None:
            self._predictor = predictor
Nikita Titov's avatar
Nikita Titov committed
1129
            return self._free_handle()
Guolin Ke's avatar
Guolin Ke committed
1130
        else:
1131
            raise LightGBMError("Cannot set predictor after freed raw data, set free_raw_data=False when construct Dataset to avoid this.")
Guolin Ke's avatar
Guolin Ke committed
1132
1133

    def set_reference(self, reference):
1134
        """Set reference Dataset.
Guolin Ke's avatar
Guolin Ke committed
1135
1136
1137
1138

        Parameters
        ----------
        reference : Dataset
1139
            Reference that is used as a template to consturct the current Dataset.
Nikita Titov's avatar
Nikita Titov committed
1140
1141
1142
1143
1144

        Returns
        -------
        self : Dataset
            Dataset with set reference.
Guolin Ke's avatar
Guolin Ke committed
1145
        """
Nikita Titov's avatar
Nikita Titov committed
1146
        self.set_categorical_feature(reference.categorical_feature).set_feature_name(reference.feature_name)._set_predictor(reference._predictor)
1147
1148
        # 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
1149
            return self
Guolin Ke's avatar
Guolin Ke committed
1150
1151
        if self.data is not None:
            self.reference = reference
Nikita Titov's avatar
Nikita Titov committed
1152
            return self._free_handle()
Guolin Ke's avatar
Guolin Ke committed
1153
        else:
1154
            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
1155
1156

    def set_feature_name(self, feature_name):
1157
        """Set feature name.
Guolin Ke's avatar
Guolin Ke committed
1158
1159
1160

        Parameters
        ----------
1161
1162
        feature_name : list of strings
            Feature names.
Nikita Titov's avatar
Nikita Titov committed
1163
1164
1165
1166
1167

        Returns
        -------
        self : Dataset
            Dataset with set feature name.
Guolin Ke's avatar
Guolin Ke committed
1168
        """
1169
1170
        if feature_name != 'auto':
            self.feature_name = feature_name
1171
        if self.handle is not None and feature_name is not None and feature_name != 'auto':
wxchan's avatar
wxchan committed
1172
1173
            if len(feature_name) != self.num_feature():
                raise ValueError("Length of feature_name({}) and num_feature({}) don't match".format(len(feature_name), self.num_feature()))
1174
            c_feature_name = [c_str(name) for name in feature_name]
wxchan's avatar
wxchan committed
1175
1176
1177
            _safe_call(_LIB.LGBM_DatasetSetFeatureNames(
                self.handle,
                c_array(ctypes.c_char_p, c_feature_name),
Guolin Ke's avatar
Guolin Ke committed
1178
                ctypes.c_int(len(feature_name))))
Nikita Titov's avatar
Nikita Titov committed
1179
        return self
Guolin Ke's avatar
Guolin Ke committed
1180
1181

    def set_label(self, label):
1182
        """Set label of Dataset
Guolin Ke's avatar
Guolin Ke committed
1183
1184
1185

        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
1186
        label : list, numpy array or None
1187
            The label information to be set into Dataset.
Nikita Titov's avatar
Nikita Titov committed
1188
1189
1190
1191
1192

        Returns
        -------
        self : Dataset
            Dataset with set label.
Guolin Ke's avatar
Guolin Ke committed
1193
1194
        """
        self.label = label
1195
        if self.handle is not None:
wxchan's avatar
wxchan committed
1196
1197
            label = list_to_1d_numpy(label, name='label')
            self.set_field('label', label)
Nikita Titov's avatar
Nikita Titov committed
1198
        return self
Guolin Ke's avatar
Guolin Ke committed
1199
1200

    def set_weight(self, weight):
1201
        """Set weight of each instance.
Guolin Ke's avatar
Guolin Ke committed
1202
1203
1204

        Parameters
        ----------
1205
1206
        weight : list, numpy array or None
            Weight to be set for each data point.
Nikita Titov's avatar
Nikita Titov committed
1207
1208
1209
1210
1211

        Returns
        -------
        self : Dataset
            Dataset with set weight.
Guolin Ke's avatar
Guolin Ke committed
1212
        """
1213
1214
        if weight is not None and np.all(weight == 1):
            weight = None
Guolin Ke's avatar
Guolin Ke committed
1215
        self.weight = weight
1216
        if self.handle is not None and weight is not None:
wxchan's avatar
wxchan committed
1217
1218
            weight = list_to_1d_numpy(weight, name='weight')
            self.set_field('weight', weight)
Nikita Titov's avatar
Nikita Titov committed
1219
        return self
Guolin Ke's avatar
Guolin Ke committed
1220
1221

    def set_init_score(self, init_score):
1222
        """Set init score of Booster to start from.
Guolin Ke's avatar
Guolin Ke committed
1223
1224
1225

        Parameters
        ----------
1226
1227
        init_score : list, numpy array or None
            Init score for Booster.
Nikita Titov's avatar
Nikita Titov committed
1228
1229
1230
1231
1232

        Returns
        -------
        self : Dataset
            Dataset with set init score.
Guolin Ke's avatar
Guolin Ke committed
1233
1234
        """
        self.init_score = init_score
1235
        if self.handle is not None and init_score is not None:
Guolin Ke's avatar
Guolin Ke committed
1236
            init_score = list_to_1d_numpy(init_score, np.float64, name='init_score')
wxchan's avatar
wxchan committed
1237
            self.set_field('init_score', init_score)
Nikita Titov's avatar
Nikita Titov committed
1238
        return self
Guolin Ke's avatar
Guolin Ke committed
1239
1240

    def set_group(self, group):
1241
        """Set group size of Dataset (used for ranking).
Guolin Ke's avatar
Guolin Ke committed
1242
1243
1244

        Parameters
        ----------
1245
1246
        group : list, numpy array or None
            Group size of each group.
Nikita Titov's avatar
Nikita Titov committed
1247
1248
1249
1250
1251

        Returns
        -------
        self : Dataset
            Dataset with set group.
Guolin Ke's avatar
Guolin Ke committed
1252
1253
        """
        self.group = group
1254
        if self.handle is not None and group is not None:
wxchan's avatar
wxchan committed
1255
1256
            group = list_to_1d_numpy(group, np.int32, name='group')
            self.set_field('group', group)
Nikita Titov's avatar
Nikita Titov committed
1257
        return self
Guolin Ke's avatar
Guolin Ke committed
1258
1259

    def get_label(self):
1260
        """Get the label of the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1261
1262
1263

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1264
        label : numpy array or None
1265
            The label information from the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1266
        """
1267
        if self.label is None:
wxchan's avatar
wxchan committed
1268
            self.label = self.get_field('label')
Guolin Ke's avatar
Guolin Ke committed
1269
1270
1271
        return self.label

    def get_weight(self):
1272
        """Get the weight of the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1273
1274
1275

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1276
        weight : numpy array or None
1277
            Weight for each data point from the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1278
        """
1279
        if self.weight is None:
wxchan's avatar
wxchan committed
1280
            self.weight = self.get_field('weight')
Guolin Ke's avatar
Guolin Ke committed
1281
1282
1283
        return self.weight

    def get_init_score(self):
1284
        """Get the initial score of the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1285
1286
1287

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1288
        init_score : numpy array or None
1289
            Init score of Booster.
Guolin Ke's avatar
Guolin Ke committed
1290
        """
1291
        if self.init_score is None:
wxchan's avatar
wxchan committed
1292
            self.init_score = self.get_field('init_score')
Guolin Ke's avatar
Guolin Ke committed
1293
1294
1295
        return self.init_score

    def get_group(self):
1296
        """Get the group of the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1297
1298
1299

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1300
        group : numpy array or None
1301
            Group size of each group.
Guolin Ke's avatar
Guolin Ke committed
1302
        """
1303
        if self.group is None:
wxchan's avatar
wxchan committed
1304
            self.group = self.get_field('group')
Guolin Ke's avatar
Guolin Ke committed
1305
1306
            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
1307
                self.group = np.diff(self.group)
Guolin Ke's avatar
Guolin Ke committed
1308
1309
1310
        return self.group

    def num_data(self):
1311
        """Get the number of rows in the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1312
1313
1314

        Returns
        -------
1315
1316
        number_of_rows : int
            The number of rows in the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1317
        """
1318
        if self.handle is not None:
Guolin Ke's avatar
Guolin Ke committed
1319
            ret = ctypes.c_int()
wxchan's avatar
wxchan committed
1320
1321
1322
            _safe_call(_LIB.LGBM_DatasetGetNumData(self.handle,
                                                   ctypes.byref(ret)))
            return ret.value
Guolin Ke's avatar
Guolin Ke committed
1323
        else:
1324
            raise LightGBMError("Cannot get num_data before construct dataset")
Guolin Ke's avatar
Guolin Ke committed
1325
1326

    def num_feature(self):
1327
        """Get the number of columns (features) in the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1328
1329
1330

        Returns
        -------
1331
1332
        number_of_columns : int
            The number of columns (features) in the Dataset.
Guolin Ke's avatar
Guolin Ke committed
1333
        """
1334
        if self.handle is not None:
Guolin Ke's avatar
Guolin Ke committed
1335
            ret = ctypes.c_int()
wxchan's avatar
wxchan committed
1336
1337
1338
            _safe_call(_LIB.LGBM_DatasetGetNumFeature(self.handle,
                                                      ctypes.byref(ret)))
            return ret.value
Guolin Ke's avatar
Guolin Ke committed
1339
        else:
1340
            raise LightGBMError("Cannot get num_feature before construct dataset")
Guolin Ke's avatar
Guolin Ke committed
1341

1342
    def get_ref_chain(self, ref_limit=100):
1343
1344
1345
1346
1347
1348
1349
        """Get a chain of Dataset objects, starting with r, then going to r.reference if exists,
        then to r.reference.reference, etc. until we hit ``ref_limit`` or a reference loop.

        Parameters
        ----------
        ref_limit : int, optional (default=100)
            The limit number of references.
1350
1351
1352

        Returns
        -------
1353
1354
1355
        ref_chain : set of Dataset
            Chain of references of the Datasets.
        """
1356
        head = self
1357
        ref_chain = set()
1358
1359
        while len(ref_chain) < ref_limit:
            if isinstance(head, Dataset):
1360
                ref_chain.add(head)
1361
1362
1363
1364
1365
1366
                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
1367
        return ref_chain
1368

wxchan's avatar
wxchan committed
1369

wxchan's avatar
wxchan committed
1370
class Booster(object):
1371
    """Booster in LightGBM."""
wxchan's avatar
wxchan committed
1372
    def __init__(self, params=None, train_set=None, model_file=None, silent=False):
1373
        """Initialize the Booster.
wxchan's avatar
wxchan committed
1374
1375
1376

        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
1377
        params : dict or None, optional (default=None)
1378
1379
1380
1381
            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
1382
            Path to the model file.
1383
1384
        silent : bool, optional (default=False)
            Whether to print messages during construction.
wxchan's avatar
wxchan committed
1385
        """
1386
        self.handle = None
1387
        self.network = False
wxchan's avatar
wxchan committed
1388
1389
1390
        self.__need_reload_eval_info = True
        self.__train_data_name = "training"
        self.__attr = {}
1391
        self.__set_objective_to_none = False
wxchan's avatar
wxchan committed
1392
        self.best_iteration = -1
wxchan's avatar
wxchan committed
1393
        self.best_score = {}
wxchan's avatar
wxchan committed
1394
1395
1396
1397
        params = {} if params is None else params
        if silent:
            params["verbose"] = 0
        if train_set is not None:
1398
            # Training task
wxchan's avatar
wxchan committed
1399
            if not isinstance(train_set, Dataset):
1400
                raise TypeError('Training data should be Dataset instance, met {}'.format(type(train_set).__name__))
wxchan's avatar
wxchan committed
1401
            params_str = param_dict_to_str(params)
1402
            # construct booster object
1403
            self.handle = ctypes.c_void_p()
wxchan's avatar
wxchan committed
1404
            _safe_call(_LIB.LGBM_BoosterCreate(
wxchan's avatar
wxchan committed
1405
                train_set.construct().handle,
wxchan's avatar
wxchan committed
1406
1407
                c_str(params_str),
                ctypes.byref(self.handle)))
1408
            # save reference to data
wxchan's avatar
wxchan committed
1409
1410
1411
1412
            self.train_set = train_set
            self.valid_sets = []
            self.name_valid_sets = []
            self.__num_dataset = 1
Guolin Ke's avatar
Guolin Ke committed
1413
1414
            self.__init_predictor = train_set._predictor
            if self.__init_predictor is not None:
wxchan's avatar
wxchan committed
1415
1416
                _safe_call(_LIB.LGBM_BoosterMerge(
                    self.handle,
Guolin Ke's avatar
Guolin Ke committed
1417
                    self.__init_predictor.handle))
Guolin Ke's avatar
Guolin Ke committed
1418
            out_num_class = ctypes.c_int(0)
wxchan's avatar
wxchan committed
1419
1420
1421
1422
            _safe_call(_LIB.LGBM_BoosterGetNumClasses(
                self.handle,
                ctypes.byref(out_num_class)))
            self.__num_class = out_num_class.value
1423
            # buffer for inner predict
wxchan's avatar
wxchan committed
1424
1425
1426
            self.__inner_predict_buffer = [None]
            self.__is_predicted_cur_iter = [False]
            self.__get_eval_info()
1427
            self.pandas_categorical = train_set.pandas_categorical
1428
            # set network if necessary
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
            for alias in ["machines", "workers", "nodes"]:
                if alias in params:
                    machines = params[alias]
                    if isinstance(machines, string_type):
                        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),
                                     num_machines=params.get("num_machines", num_machines))
                    break
wxchan's avatar
wxchan committed
1444
        elif model_file is not None:
1445
            # Prediction task
Guolin Ke's avatar
Guolin Ke committed
1446
            out_num_iterations = ctypes.c_int(0)
1447
            self.handle = ctypes.c_void_p()
wxchan's avatar
wxchan committed
1448
1449
1450
1451
            _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
1452
            out_num_class = ctypes.c_int(0)
wxchan's avatar
wxchan committed
1453
1454
1455
1456
            _safe_call(_LIB.LGBM_BoosterGetNumClasses(
                self.handle,
                ctypes.byref(out_num_class)))
            self.__num_class = out_num_class.value
1457
            self.pandas_categorical = _load_pandas_categorical(model_file)
1458
        elif 'model_str' in params:
1459
            self.model_from_string(params['model_str'])
wxchan's avatar
wxchan committed
1460
        else:
1461
            raise TypeError('Need at least one training dataset or model file to create booster instance')
Guolin Ke's avatar
Guolin Ke committed
1462
        self.params = params.copy()
wxchan's avatar
wxchan committed
1463
1464

    def __del__(self):
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
        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
1475

wxchan's avatar
wxchan committed
1476
1477
1478
1479
    def __copy__(self):
        return self.__deepcopy__(None)

    def __deepcopy__(self, _):
1480
        model_str = self.model_to_string(num_iteration=-1)
1481
        booster = Booster({'model_str': model_str})
1482
        booster.pandas_categorical = self.pandas_categorical
1483
        return booster
wxchan's avatar
wxchan committed
1484
1485
1486
1487
1488
1489
1490

    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:
1491
            this["handle"] = self.model_to_string(num_iteration=-1)
wxchan's avatar
wxchan committed
1492
1493
1494
        return this

    def __setstate__(self, state):
1495
1496
        model_str = state.get('handle', None)
        if model_str is not None:
wxchan's avatar
wxchan committed
1497
            handle = ctypes.c_void_p()
Guolin Ke's avatar
Guolin Ke committed
1498
            out_num_iterations = ctypes.c_int(0)
1499
1500
1501
1502
            _safe_call(_LIB.LGBM_BoosterLoadModelFromString(
                c_str(model_str),
                ctypes.byref(out_num_iterations),
                ctypes.byref(handle)))
wxchan's avatar
wxchan committed
1503
1504
1505
            state['handle'] = handle
        self.__dict__.update(state)

wxchan's avatar
wxchan committed
1506
    def free_dataset(self):
Nikita Titov's avatar
Nikita Titov committed
1507
1508
1509
1510
1511
1512
1513
        """Free Booster's Datasets.

        Returns
        -------
        self : Booster
            Booster without Datasets.
        """
wxchan's avatar
wxchan committed
1514
1515
        self.__dict__.pop('train_set', None)
        self.__dict__.pop('valid_sets', None)
1516
        self.__num_dataset = 0
Nikita Titov's avatar
Nikita Titov committed
1517
        return self
wxchan's avatar
wxchan committed
1518

1519
1520
1521
    def _free_buffer(self):
        self.__inner_predict_buffer = []
        self.__is_predicted_cur_iter = []
Nikita Titov's avatar
Nikita Titov committed
1522
        return self
1523

1524
1525
1526
1527
1528
1529
    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
1530
        machines : list, set or string
1531
            Names of machines.
Nikita Titov's avatar
Nikita Titov committed
1532
        local_listen_port : int, optional (default=12400)
1533
            TCP listen port for local machines.
Nikita Titov's avatar
Nikita Titov committed
1534
        listen_time_out : int, optional (default=120)
1535
            Socket time-out in minutes.
Nikita Titov's avatar
Nikita Titov committed
1536
        num_machines : int, optional (default=1)
1537
            The number of machines for parallel learning application.
Nikita Titov's avatar
Nikita Titov committed
1538
1539
1540
1541
1542

        Returns
        -------
        self : Booster
            Booster with set network.
1543
1544
1545
1546
1547
1548
        """
        _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
1549
        return self
1550
1551

    def free_network(self):
Nikita Titov's avatar
Nikita Titov committed
1552
1553
1554
1555
1556
1557
1558
        """Free Booster's network.

        Returns
        -------
        self : Booster
            Booster with freed network.
        """
1559
1560
        _safe_call(_LIB.LGBM_NetworkFree())
        self.network = False
Nikita Titov's avatar
Nikita Titov committed
1561
        return self
1562

wxchan's avatar
wxchan committed
1563
    def set_train_data_name(self, name):
1564
1565
1566
1567
        """Set the name to the training Dataset.

        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
1568
1569
1570
1571
1572
1573
1574
        name : string
            Name for the training Dataset.

        Returns
        -------
        self : Booster
            Booster with set training Dataset name.
1575
        """
wxchan's avatar
wxchan committed
1576
        self.__train_data_name = name
Nikita Titov's avatar
Nikita Titov committed
1577
        return self
wxchan's avatar
wxchan committed
1578
1579

    def add_valid(self, data, name):
1580
        """Add validation data.
wxchan's avatar
wxchan committed
1581
1582
1583
1584

        Parameters
        ----------
        data : Dataset
1585
1586
1587
            Validation data.
        name : string
            Name of validation data.
Nikita Titov's avatar
Nikita Titov committed
1588
1589
1590
1591
1592

        Returns
        -------
        self : Booster
            Booster with set validation data.
wxchan's avatar
wxchan committed
1593
        """
Guolin Ke's avatar
Guolin Ke committed
1594
        if not isinstance(data, Dataset):
1595
            raise TypeError('Validation data should be Dataset instance, met {}'.format(type(data).__name__))
Guolin Ke's avatar
Guolin Ke committed
1596
1597
        if data._predictor is not self.__init_predictor:
            raise LightGBMError("Add validation data failed, you should use same predictor for these data")
wxchan's avatar
wxchan committed
1598
1599
        _safe_call(_LIB.LGBM_BoosterAddValidData(
            self.handle,
wxchan's avatar
wxchan committed
1600
            data.construct().handle))
wxchan's avatar
wxchan committed
1601
1602
1603
1604
1605
        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
1606
        return self
wxchan's avatar
wxchan committed
1607
1608

    def reset_parameter(self, params):
1609
        """Reset parameters of Booster.
wxchan's avatar
wxchan committed
1610
1611
1612
1613

        Parameters
        ----------
        params : dict
1614
            New parameters for Booster.
Nikita Titov's avatar
Nikita Titov committed
1615
1616
1617
1618
1619

        Returns
        -------
        self : Booster
            Booster with new parameters.
wxchan's avatar
wxchan committed
1620
        """
1621
        if any(metric_alias in params for metric_alias in ('metric', 'metrics', 'metric_types')):
wxchan's avatar
wxchan committed
1622
1623
1624
1625
1626
1627
            self.__need_reload_eval_info = True
        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
1628
        self.params.update(params)
Nikita Titov's avatar
Nikita Titov committed
1629
        return self
wxchan's avatar
wxchan committed
1630
1631

    def update(self, train_set=None, fobj=None):
Nikita Titov's avatar
Nikita Titov committed
1632
        """Update Booster for one iteration.
1633

wxchan's avatar
wxchan committed
1634
1635
        Parameters
        ----------
1636
1637
1638
1639
        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
1640
1641
            Customized objective function.

1642
1643
1644
1645
            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.

wxchan's avatar
wxchan committed
1646
1647
        Returns
        -------
1648
1649
        is_finished : bool
            Whether the update was successfully finished.
wxchan's avatar
wxchan committed
1650
1651
        """

1652
        # need reset training data
wxchan's avatar
wxchan committed
1653
        if train_set is not None and train_set is not self.train_set:
Guolin Ke's avatar
Guolin Ke committed
1654
1655
            if not isinstance(train_set, Dataset):
                raise TypeError('Training data should be Dataset instance, met {}'.format(type(train_set).__name__))
Guolin Ke's avatar
Guolin Ke committed
1656
1657
            if train_set._predictor is not self.__init_predictor:
                raise LightGBMError("Replace training data failed, you should use same predictor for these data")
wxchan's avatar
wxchan committed
1658
1659
1660
            self.train_set = train_set
            _safe_call(_LIB.LGBM_BoosterResetTrainingData(
                self.handle,
wxchan's avatar
wxchan committed
1661
                self.train_set.construct().handle))
wxchan's avatar
wxchan committed
1662
1663
1664
            self.__inner_predict_buffer[0] = None
        is_finished = ctypes.c_int(0)
        if fobj is None:
1665
1666
            if self.__set_objective_to_none:
                raise ValueError('Cannot update due to null objective function.')
wxchan's avatar
wxchan committed
1667
1668
1669
            _safe_call(_LIB.LGBM_BoosterUpdateOneIter(
                self.handle,
                ctypes.byref(is_finished)))
wxchan's avatar
wxchan committed
1670
            self.__is_predicted_cur_iter = [False for _ in range_(self.__num_dataset)]
wxchan's avatar
wxchan committed
1671
1672
            return is_finished.value == 1
        else:
1673
            if not self.__set_objective_to_none:
Nikita Titov's avatar
Nikita Titov committed
1674
                self.reset_parameter({"objective": "none"}).__set_objective_to_none = True
wxchan's avatar
wxchan committed
1675
1676
1677
1678
1679
            grad, hess = fobj(self.__inner_predict(0), self.train_set)
            return self.__boost(grad, hess)

    def __boost(self, grad, hess):
        """
Nikita Titov's avatar
Nikita Titov committed
1680
1681
1682
1683
1684
        Boost Booster for one iteration with customized gradient statistics.

        Note: 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.
1685

wxchan's avatar
wxchan committed
1686
1687
        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
1688
1689
        grad : 1d numpy array or list
            The first order derivative (gradient).
wxchan's avatar
wxchan committed
1690
        hess : 1d numpy or 1d list
Nikita Titov's avatar
Nikita Titov committed
1691
            The second order derivative (Hessian).
wxchan's avatar
wxchan committed
1692
1693
1694

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1695
1696
        is_finished : bool
            Whether the boost was successfully finished.
wxchan's avatar
wxchan committed
1697
        """
1698
1699
        grad = list_to_1d_numpy(grad, name='gradient')
        hess = list_to_1d_numpy(hess, name='hessian')
1700
1701
        assert grad.flags.c_contiguous
        assert hess.flags.c_contiguous
wxchan's avatar
wxchan committed
1702
        if len(grad) != len(hess):
1703
            raise ValueError("Lengths of gradient({}) and hessian({}) don't match".format(len(grad), len(hess)))
wxchan's avatar
wxchan committed
1704
1705
1706
1707
1708
1709
        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)))
wxchan's avatar
wxchan committed
1710
        self.__is_predicted_cur_iter = [False for _ in range_(self.__num_dataset)]
wxchan's avatar
wxchan committed
1711
1712
1713
        return is_finished.value == 1

    def rollback_one_iter(self):
Nikita Titov's avatar
Nikita Titov committed
1714
1715
1716
1717
1718
1719
1720
        """Rollback one iteration.

        Returns
        -------
        self : Booster
            Booster with rolled back one iteration.
        """
wxchan's avatar
wxchan committed
1721
1722
        _safe_call(_LIB.LGBM_BoosterRollbackOneIter(
            self.handle))
wxchan's avatar
wxchan committed
1723
        self.__is_predicted_cur_iter = [False for _ in range_(self.__num_dataset)]
Nikita Titov's avatar
Nikita Titov committed
1724
        return self
wxchan's avatar
wxchan committed
1725
1726

    def current_iteration(self):
1727
1728
1729
1730
1731
1732
1733
        """Get the index of the current iteration.

        Returns
        -------
        cur_iter : int
            The index of the current iteration.
        """
Guolin Ke's avatar
Guolin Ke committed
1734
        out_cur_iter = ctypes.c_int(0)
wxchan's avatar
wxchan committed
1735
1736
1737
1738
1739
1740
        _safe_call(_LIB.LGBM_BoosterGetCurrentIteration(
            self.handle,
            ctypes.byref(out_cur_iter)))
        return out_cur_iter.value

    def eval(self, data, name, feval=None):
1741
        """Evaluate for data.
wxchan's avatar
wxchan committed
1742
1743
1744

        Parameters
        ----------
1745
1746
1747
1748
1749
        data : Dataset
            Data for the evaluating.
        name : string
            Name of the data.
        feval : callable or None, optional (default=None)
1750
1751
1752
1753
1754
            Customized evaluation function.
            Should accept two parameters: preds, train_data.
            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].
            Note: should return (eval_name, eval_result, is_higher_better) or list of such tuples.
1755

wxchan's avatar
wxchan committed
1756
1757
        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1758
        result : list
1759
            List with evaluation results.
wxchan's avatar
wxchan committed
1760
        """
Guolin Ke's avatar
Guolin Ke committed
1761
1762
        if not isinstance(data, Dataset):
            raise TypeError("Can only eval for Dataset instance")
wxchan's avatar
wxchan committed
1763
1764
1765
1766
        data_idx = -1
        if data is self.train_set:
            data_idx = 0
        else:
wxchan's avatar
wxchan committed
1767
            for i in range_(len(self.valid_sets)):
wxchan's avatar
wxchan committed
1768
1769
1770
                if data is self.valid_sets[i]:
                    data_idx = i + 1
                    break
1771
        # need to push new valid data
wxchan's avatar
wxchan committed
1772
1773
1774
1775
1776
1777
1778
        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):
1779
        """Evaluate for training data.
wxchan's avatar
wxchan committed
1780
1781
1782

        Parameters
        ----------
1783
        feval : callable or None, optional (default=None)
1784
1785
1786
1787
1788
            Customized evaluation function.
            Should accept two parameters: preds, train_data.
            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].
            Note: should return (eval_name, eval_result, is_higher_better) or list of such tuples.
wxchan's avatar
wxchan committed
1789
1790
1791

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1792
        result : list
1793
            List with evaluation results.
wxchan's avatar
wxchan committed
1794
1795
1796
1797
        """
        return self.__inner_eval(self.__train_data_name, 0, feval)

    def eval_valid(self, feval=None):
1798
        """Evaluate for validation data.
wxchan's avatar
wxchan committed
1799
1800
1801

        Parameters
        ----------
1802
        feval : callable or None, optional (default=None)
1803
1804
1805
1806
1807
            Customized evaluation function.
            Should accept two parameters: preds, train_data.
            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].
            Note: should return (eval_name, eval_result, is_higher_better) or list of such tuples.
wxchan's avatar
wxchan committed
1808
1809
1810

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1811
        result : list
1812
            List with evaluation results.
wxchan's avatar
wxchan committed
1813
        """
wxchan's avatar
wxchan committed
1814
        return [item for i in range_(1, self.__num_dataset)
wxchan's avatar
wxchan committed
1815
                for item in self.__inner_eval(self.name_valid_sets[i - 1], i, feval)]
wxchan's avatar
wxchan committed
1816

1817
    def save_model(self, filename, num_iteration=None, start_iteration=0):
1818
        """Save Booster to file.
wxchan's avatar
wxchan committed
1819
1820
1821

        Parameters
        ----------
1822
1823
        filename : string
            Filename to save Booster.
1824
1825
1826
1827
        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
1828
        start_iteration : int, optional (default=0)
1829
            Start index of the iteration that should be saved.
Nikita Titov's avatar
Nikita Titov committed
1830
1831
1832
1833
1834

        Returns
        -------
        self : Booster
            Returns self.
wxchan's avatar
wxchan committed
1835
        """
1836
        if num_iteration is None:
1837
            num_iteration = self.best_iteration
wxchan's avatar
wxchan committed
1838
1839
        _safe_call(_LIB.LGBM_BoosterSaveModel(
            self.handle,
1840
            ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
1841
            ctypes.c_int(num_iteration),
wxchan's avatar
wxchan committed
1842
            c_str(filename)))
1843
        _save_pandas_categorical(filename, self.pandas_categorical)
Nikita Titov's avatar
Nikita Titov committed
1844
        return self
wxchan's avatar
wxchan committed
1845

1846
1847
    def shuffle_models(self):
        """Shuffle models.
Nikita Titov's avatar
Nikita Titov committed
1848
1849
1850
1851
1852

        Returns
        -------
        self : Booster
            Booster with shuffled models.
1853
1854
        """
        _safe_call(_LIB.LGBM_BoosterShuffleModels(self.handle))
Nikita Titov's avatar
Nikita Titov committed
1855
        return self
1856
1857
1858
1859
1860
1861

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

        Parameters
        ----------
Nikita Titov's avatar
Nikita Titov committed
1862
        model_str : string
1863
            Model will be loaded from this string.
Nikita Titov's avatar
Nikita Titov committed
1864
1865
        verbose : bool, optional (default=True)
            Whether to print messages while loading model.
1866
1867
1868

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1869
        self : Booster
1870
1871
            Loaded Booster object.
        """
1872
1873
1874
1875
        if self.handle is not None:
            _safe_call(_LIB.LGBM_BoosterFree(self.handle))
        self._free_buffer()
        self.handle = ctypes.c_void_p()
1876
1877
1878
1879
1880
1881
1882
1883
1884
        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)))
1885
        if verbose:
Nikita Titov's avatar
Nikita Titov committed
1886
            print('Finished loading model, total used %d iterations' % int(out_num_iterations.value))
1887
        self.__num_class = out_num_class.value
1888
1889
1890
1891
        return self

    def model_to_string(self, num_iteration=None, start_iteration=0):
        """Save Booster to string.
1892

1893
1894
1895
1896
1897
1898
        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
1899
        start_iteration : int, optional (default=0)
1900
1901
1902
1903
            Start index of the iteration that should be saved.

        Returns
        -------
Nikita Titov's avatar
Nikita Titov committed
1904
        str_repr : string
1905
1906
            String representation of Booster.
        """
1907
        if num_iteration is None:
1908
1909
            num_iteration = self.best_iteration
        buffer_len = 1 << 20
1910
        tmp_out_len = ctypes.c_int64(0)
1911
1912
1913
1914
        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,
1915
            ctypes.c_int(start_iteration),
1916
            ctypes.c_int(num_iteration),
1917
            ctypes.c_int64(buffer_len),
1918
1919
1920
1921
1922
1923
1924
1925
1926
            ctypes.byref(tmp_out_len),
            ptr_string_buffer))
        actual_len = tmp_out_len.value
        '''if buffer length is not long enough, re-allocate a buffer'''
        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,
1927
                ctypes.c_int(start_iteration),
1928
                ctypes.c_int(num_iteration),
1929
                ctypes.c_int64(actual_len),
1930
1931
1932
1933
                ctypes.byref(tmp_out_len),
                ptr_string_buffer))
        return string_buffer.value.decode()

1934
    def dump_model(self, num_iteration=None, start_iteration=0):
Nikita Titov's avatar
Nikita Titov committed
1935
        """Dump Booster to JSON format.
wxchan's avatar
wxchan committed
1936

1937
1938
        Parameters
        ----------
1939
1940
1941
1942
        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
1943
        start_iteration : int, optional (default=0)
1944
            Start index of the iteration that should be dumped.
1945

wxchan's avatar
wxchan committed
1946
1947
        Returns
        -------
1948
        json_repr : dict
Nikita Titov's avatar
Nikita Titov committed
1949
            JSON format of Booster.
wxchan's avatar
wxchan committed
1950
        """
1951
        if num_iteration is None:
1952
            num_iteration = self.best_iteration
wxchan's avatar
wxchan committed
1953
        buffer_len = 1 << 20
1954
        tmp_out_len = ctypes.c_int64(0)
wxchan's avatar
wxchan committed
1955
1956
1957
1958
        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,
1959
            ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
1960
            ctypes.c_int(num_iteration),
1961
            ctypes.c_int64(buffer_len),
wxchan's avatar
wxchan committed
1962
            ctypes.byref(tmp_out_len),
Guolin Ke's avatar
Guolin Ke committed
1963
            ptr_string_buffer))
wxchan's avatar
wxchan committed
1964
        actual_len = tmp_out_len.value
1965
        '''if buffer length is not long enough, reallocate a buffer'''
wxchan's avatar
wxchan committed
1966
1967
1968
1969
1970
        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,
1971
                ctypes.c_int(start_iteration),
Guolin Ke's avatar
Guolin Ke committed
1972
                ctypes.c_int(num_iteration),
1973
                ctypes.c_int64(actual_len),
wxchan's avatar
wxchan committed
1974
                ctypes.byref(tmp_out_len),
Guolin Ke's avatar
Guolin Ke committed
1975
                ptr_string_buffer))
wxchan's avatar
wxchan committed
1976
1977
        return json.loads(string_buffer.value.decode())

1978
    def predict(self, data, num_iteration=None, raw_score=False, pred_leaf=False, pred_contrib=False,
1979
                data_has_header=False, is_reshape=True, pred_parameter=None, **kwargs):
1980
        """Make a prediction.
wxchan's avatar
wxchan committed
1981
1982
1983

        Parameters
        ----------
1984
1985
1986
        data : string, numpy array or scipy.sparse
            Data source for prediction.
            If string, it represents the path to txt file.
1987
1988
1989
1990
1991
        num_iteration : int or None, optional (default=None)
            Limit number of iterations in the prediction.
            If None, if the best iteration exists, it is used; otherwise, all iterations are used.
            If <= 0, all iterations are used (no limits).

1992
1993
1994
1995
        raw_score : bool, optional (default=False)
            Whether to predict raw scores.
        pred_leaf : bool, optional (default=False)
            Whether to predict leaf index.
1996
1997
        pred_contrib : bool, optional (default=False)
            Whether to predict feature contributions.
1998
1999
2000
2001
2002
        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].
2003
2004
        pred_parameter : dict or None, optional (default=None)
            Deprecated.
2005
            Other parameters for the prediction.
2006
        **kwargs : other parameters for the prediction
wxchan's avatar
wxchan committed
2007
2008
2009

        Returns
        -------
2010
2011
        result : numpy array
            Prediction result.
wxchan's avatar
wxchan committed
2012
        """
2013
2014
2015
2016
2017
2018
        if pred_parameter:
            warnings.warn("pred_parameter is deprecated and will be removed in 2.2 version.\n"
                          "Please use kwargs instead.", LGBMDeprecationWarning)
            pred_parameter.update(kwargs)
        else:
            pred_parameter = kwargs
2019
        predictor = self._to_predictor(pred_parameter)
2020
        if num_iteration is None:
2021
            num_iteration = self.best_iteration
2022
        return predictor.predict(data, num_iteration, raw_score, pred_leaf, pred_contrib, data_has_header, is_reshape)
wxchan's avatar
wxchan committed
2023

Guolin Ke's avatar
Guolin Ke committed
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
    def refit(self, data, label, decay_rate=0.9):
        """Refit the existing Booster by new data.

        Parameters
        ----------
        data : string, numpy array or scipy.sparse
            Data source for refit.
            If string, it represents the path to txt file.
        label : list or numpy 1-D array
            Label for refit.
        decay_rate : float, optional (default=0.9)
            Decay rate of refit, will use ``leaf_output = decay_rate * old_leaf_output + (1.0 - decay_rate) * new_leaf_output`` to refit trees.

        Returns
        -------
        result : Booster
            Refitted Booster.
        """
        predictor = self._to_predictor()
        leaf_preds = predictor.predict(data, -1, pred_leaf=True)
        nrow = leaf_preds.shape[0]
        ncol = leaf_preds.shape[1]
        train_set = Dataset(data, label)
        new_booster = Booster(self.params, train_set, silent=True)
        # Copy models
        _safe_call(_LIB.LGBM_BoosterMerge(
            new_booster.handle,
            predictor.handle))
        leaf_preds = leaf_preds.reshape(-1)
        ptr_data, type_ptr_data, _ = c_int_array(leaf_preds)
        _safe_call(_LIB.LGBM_BoosterRefit(
            new_booster.handle,
            ptr_data,
            ctypes.c_int(nrow),
            ctypes.c_int(ncol)))
        return new_booster

2061
    def get_leaf_output(self, tree_id, leaf_id):
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
        """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.
        """
2076
2077
2078
2079
2080
2081
2082
2083
        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

2084
    def _to_predictor(self, pred_parameter=None):
wxchan's avatar
wxchan committed
2085
        """Convert to predictor"""
2086
        predictor = _InnerPredictor(booster_handle=self.handle, pred_parameter=pred_parameter)
2087
        predictor.pandas_categorical = self.pandas_categorical
wxchan's avatar
wxchan committed
2088
2089
        return predictor

2090
    def num_feature(self):
2091
2092
2093
2094
2095
2096
2097
        """Get number of features.

        Returns
        -------
        num_feature : int
            The number of features.
        """
2098
2099
2100
2101
2102
2103
        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
2104
    def feature_name(self):
2105
        """Get names of features.
wxchan's avatar
wxchan committed
2106
2107
2108

        Returns
        -------
2109
2110
        result : list
            List with names of features.
wxchan's avatar
wxchan committed
2111
        """
2112
        num_feature = self.num_feature()
2113
        # Get name of features
wxchan's avatar
wxchan committed
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
        tmp_out_len = ctypes.c_int(0)
        string_buffers = [ctypes.create_string_buffer(255) for i in range_(num_feature)]
        ptr_string_buffers = (ctypes.c_char_p * num_feature)(*map(ctypes.addressof, string_buffers))
        _safe_call(_LIB.LGBM_BoosterGetFeatureNames(
            self.handle,
            ctypes.byref(tmp_out_len),
            ptr_string_buffers))
        if num_feature != tmp_out_len.value:
            raise ValueError("Length of feature names doesn't equal with num_feature")
        return [string_buffers[i].value.decode() for i in range_(num_feature)]

2125
    def feature_importance(self, importance_type='split', iteration=-1):
2126
        """Get feature importances.
2127

2128
2129
        Parameters
        ----------
2130
2131
2132
2133
        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.
2134

2135
2136
        Returns
        -------
2137
2138
        result : numpy array
            Array with feature importances.
2139
        """
2140
2141
2142
2143
2144
2145
        if importance_type == "split":
            importance_type_int = 0
        elif importance_type == "gain":
            importance_type_int = 1
        else:
            importance_type_int = -1
Nikita Titov's avatar
Nikita Titov committed
2146
        result = np.zeros(self.num_feature(), dtype=np.float64)
2147
2148
2149
2150
2151
2152
2153
2154
2155
        _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:
            return result.astype(int)
        else:
            return result
2156

wxchan's avatar
wxchan committed
2157
2158
    def __inner_eval(self, data_name, data_idx, feval=None):
        """
Nikita Titov's avatar
Nikita Titov committed
2159
        Evaluate training or validation data
wxchan's avatar
wxchan committed
2160
2161
        """
        if data_idx >= self.__num_dataset:
2162
            raise ValueError("Data_idx should be smaller than number of dataset")
wxchan's avatar
wxchan committed
2163
2164
2165
        self.__get_eval_info()
        ret = []
        if self.__num_inner_eval > 0:
2166
            result = np.zeros(self.__num_inner_eval, dtype=np.float64)
Guolin Ke's avatar
Guolin Ke committed
2167
            tmp_out_len = ctypes.c_int(0)
wxchan's avatar
wxchan committed
2168
2169
            _safe_call(_LIB.LGBM_BoosterGetEval(
                self.handle,
Guolin Ke's avatar
Guolin Ke committed
2170
                ctypes.c_int(data_idx),
wxchan's avatar
wxchan committed
2171
                ctypes.byref(tmp_out_len),
Guolin Ke's avatar
Guolin Ke committed
2172
                result.ctypes.data_as(ctypes.POINTER(ctypes.c_double))))
wxchan's avatar
wxchan committed
2173
            if tmp_out_len.value != self.__num_inner_eval:
2174
                raise ValueError("Wrong length of eval results")
wxchan's avatar
wxchan committed
2175
            for i in range_(self.__num_inner_eval):
wxchan's avatar
wxchan committed
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
                ret.append((data_name, self.__name_inner_eval[i], result[i], self.__higher_better_inner_eval[i]))
        if feval is not None:
            if data_idx == 0:
                cur_data = self.train_set
            else:
                cur_data = self.valid_sets[data_idx - 1]
            feval_ret = feval(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
                ret.append((data_name, eval_name, val, is_higher_better))
        return ret

    def __inner_predict(self, data_idx):
        """
        Predict for training and validation dataset
        """
        if data_idx >= self.__num_dataset:
2196
            raise ValueError("Data_idx should be smaller than number of dataset")
wxchan's avatar
wxchan committed
2197
2198
2199
2200
2201
        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
2202
            self.__inner_predict_buffer[data_idx] = np.zeros(n_preds, dtype=np.float64)
2203
        # avoid to predict many time in one iteration
wxchan's avatar
wxchan committed
2204
2205
        if not self.__is_predicted_cur_iter[data_idx]:
            tmp_out_len = ctypes.c_int64(0)
Guolin Ke's avatar
Guolin Ke committed
2206
            data_ptr = self.__inner_predict_buffer[data_idx].ctypes.data_as(ctypes.POINTER(ctypes.c_double))
wxchan's avatar
wxchan committed
2207
2208
            _safe_call(_LIB.LGBM_BoosterGetPredict(
                self.handle,
Guolin Ke's avatar
Guolin Ke committed
2209
                ctypes.c_int(data_idx),
wxchan's avatar
wxchan committed
2210
2211
2212
                ctypes.byref(tmp_out_len),
                data_ptr))
            if tmp_out_len.value != len(self.__inner_predict_buffer[data_idx]):
2213
                raise ValueError("Wrong length of predict results for data %d" % (data_idx))
wxchan's avatar
wxchan committed
2214
2215
2216
2217
2218
2219
2220
2221
2222
            self.__is_predicted_cur_iter[data_idx] = True
        return self.__inner_predict_buffer[data_idx]

    def __get_eval_info(self):
        """
        Get inner evaluation count and names
        """
        if self.__need_reload_eval_info:
            self.__need_reload_eval_info = False
Guolin Ke's avatar
Guolin Ke committed
2223
            out_num_eval = ctypes.c_int(0)
2224
            # Get num of inner evals
wxchan's avatar
wxchan committed
2225
2226
2227
2228
2229
            _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:
2230
                # Get name of evals
Guolin Ke's avatar
Guolin Ke committed
2231
                tmp_out_len = ctypes.c_int(0)
wxchan's avatar
wxchan committed
2232
                string_buffers = [ctypes.create_string_buffer(255) for i in range_(self.__num_inner_eval)]
wxchan's avatar
wxchan committed
2233
                ptr_string_buffers = (ctypes.c_char_p * self.__num_inner_eval)(*map(ctypes.addressof, string_buffers))
wxchan's avatar
wxchan committed
2234
2235
2236
2237
2238
                _safe_call(_LIB.LGBM_BoosterGetEvalNames(
                    self.handle,
                    ctypes.byref(tmp_out_len),
                    ptr_string_buffers))
                if self.__num_inner_eval != tmp_out_len.value:
2239
                    raise ValueError("Length of eval names doesn't equal with num_evals")
2240
                self.__name_inner_eval = \
wxchan's avatar
wxchan committed
2241
                    [string_buffers[i].value.decode() for i in range_(self.__num_inner_eval)]
2242
                self.__higher_better_inner_eval = \
2243
                    [name.startswith(('auc', 'ndcg@', 'map@')) for name in self.__name_inner_eval]
2244

wxchan's avatar
wxchan committed
2245
    def attr(self, key):
2246
        """Get attribute string from the Booster.
wxchan's avatar
wxchan committed
2247
2248
2249

        Parameters
        ----------
2250
2251
        key : string
            The name of the attribute.
wxchan's avatar
wxchan committed
2252
2253
2254

        Returns
        -------
2255
2256
        value : string or None
            The attribute value.
Nikita Titov's avatar
Nikita Titov committed
2257
            Returns None if attribute does not exist.
wxchan's avatar
wxchan committed
2258
        """
2259
        return self.__attr.get(key, None)
wxchan's avatar
wxchan committed
2260
2261

    def set_attr(self, **kwargs):
2262
        """Set the attribute of the Booster.
wxchan's avatar
wxchan committed
2263
2264
2265
2266

        Parameters
        ----------
        **kwargs
2267
2268
            The attributes to set.
            Setting a value to None deletes an attribute.
Nikita Titov's avatar
Nikita Titov committed
2269
2270
2271
2272
2273

        Returns
        -------
        self : Booster
            Booster with set attribute.
wxchan's avatar
wxchan committed
2274
2275
2276
        """
        for key, value in kwargs.items():
            if value is not None:
wxchan's avatar
wxchan committed
2277
                if not isinstance(value, string_type):
Nikita Titov's avatar
Nikita Titov committed
2278
                    raise ValueError("Only string values are accepted")
wxchan's avatar
wxchan committed
2279
2280
2281
                self.__attr[key] = value
            else:
                self.__attr.pop(key, None)
Nikita Titov's avatar
Nikita Titov committed
2282
        return self