""" The :mod:`sklearn.model_selection._split` module includes classes and functions to split the data based on a preset strategy. """ # Author: Alexandre Gramfort , # Gael Varoquaux , # Olivier Grisel # Raghav RV # License: BSD 3 clause from __future__ import print_function from __future__ import division import warnings from itertools import chain, combinations from math import ceil, floor import numbers from abc import ABCMeta, abstractmethod import numpy as np from ..utils import indexable, check_random_state, safe_indexing from ..utils.validation import _num_samples, column_or_1d from ..utils.validation import check_array from ..utils.multiclass import type_of_target from ..externals.six import with_metaclass from ..externals.six.moves import zip from ..utils.fixes import signature, comb from ..utils.fixes import _Iterable as Iterable from ..base import _pprint __all__ = ['BaseCrossValidator', 'KFold', 'GroupKFold', 'LeaveOneGroupOut', 'LeaveOneOut', 'LeavePGroupsOut', 'LeavePOut', 'RepeatedStratifiedKFold', 'RepeatedKFold', 'ShuffleSplit', 'GroupShuffleSplit', 'StratifiedKFold', 'StratifiedShuffleSplit', 'PredefinedSplit', 'train_test_split', 'check_cv'] class BaseCrossValidator(with_metaclass(ABCMeta)): """Base class for all cross-validators Implementations must define `_iter_test_masks` or `_iter_test_indices`. """ def __init__(self): # We need this for the build_repr to work properly in py2.7 # see #6304 pass def split(self, X, y=None, groups=None): """Generate indices to split data into training and test set. Parameters ---------- X : array-like, shape (n_samples, n_features) Training data, where n_samples is the number of samples and n_features is the number of features. y : array-like, of length n_samples The target variable for supervised learning problems. groups : array-like, with shape (n_samples,), optional Group labels for the samples used while splitting the dataset into train/test set. Returns ------- train : ndarray The training set indices for that split. test : ndarray The testing set indices for that split. Notes ----- Randomized CV splitters may return different results for each call of split. You can make the results identical by setting ``random_state`` to an integer. """ X, y, groups = indexable(X, y, groups) indices = np.arange(_num_samples(X)) for test_index in self._iter_test_masks(X, y, groups): train_index = indices[np.logical_not(test_index)] test_index = indices[test_index] yield train_index, test_index # Since subclasses must implement either _iter_test_masks or # _iter_test_indices, neither can be abstract. def _iter_test_masks(self, X=None, y=None, groups=None): """Generates boolean masks corresponding to test sets. By default, delegates to _iter_test_indices(X, y, groups) """ for test_index in self._iter_test_indices(X, y, groups): test_mask = np.zeros(_num_samples(X), dtype=np.bool) test_mask[test_index] = True yield test_mask def _iter_test_indices(self, X=None, y=None, groups=None): """Generates integer indices corresponding to test sets.""" raise NotImplementedError @abstractmethod def get_n_splits(self, X=None, y=None, groups=None): """Returns the number of splitting iterations in the cross-validator""" def __repr__(self): return _build_repr(self) class LeaveOneOut(BaseCrossValidator): """Leave-One-Out cross-validator Provides train/test indices to split data in train/test sets. Each sample is used once as a test set (singleton) while the remaining samples form the training set. Note: ``LeaveOneOut()`` is equivalent to ``KFold(n_splits=n)`` and ``LeavePOut(p=1)`` where ``n`` is the number of samples. Due to the high number of test sets (which is the same as the number of samples) this cross-validation method can be very costly. For large datasets one should favor :class:`KFold`, :class:`ShuffleSplit` or :class:`StratifiedKFold`. Read more in the :ref:`User Guide `. Examples -------- >>> from sklearn.model_selection import LeaveOneOut >>> X = np.array([[1, 2], [3, 4]]) >>> y = np.array([1, 2]) >>> loo = LeaveOneOut() >>> loo.get_n_splits(X) 2 >>> print(loo) LeaveOneOut() >>> for train_index, test_index in loo.split(X): ... print("TRAIN:", train_index, "TEST:", test_index) ... X_train, X_test = X[train_index], X[test_index] ... y_train, y_test = y[train_index], y[test_index] ... print(X_train, X_test, y_train, y_test) TRAIN: [1] TEST: [0] [[3 4]] [[1 2]] [2] [1] TRAIN: [0] TEST: [1] [[1 2]] [[3 4]] [1] [2] See also -------- LeaveOneGroupOut For splitting the data according to explicit, domain-specific stratification of the dataset. GroupKFold: K-fold iterator variant with non-overlapping groups. """ def _iter_test_indices(self, X, y=None, groups=None): return range(_num_samples(X)) def get_n_splits(self, X, y=None, groups=None): """Returns the number of splitting iterations in the cross-validator Parameters ---------- X : array-like, shape (n_samples, n_features) Training data, where n_samples is the number of samples and n_features is the number of features. y : object Always ignored, exists for compatibility. groups : object Always ignored, exists for compatibility. Returns ------- n_splits : int Returns the number of splitting iterations in the cross-validator. """ if X is None: raise ValueError("The 'X' parameter should not be None.") return _num_samples(X) class LeavePOut(BaseCrossValidator): """Leave-P-Out cross-validator Provides train/test indices to split data in train/test sets. This results in testing on all distinct samples of size p, while the remaining n - p samples form the training set in each iteration. Note: ``LeavePOut(p)`` is NOT equivalent to ``KFold(n_splits=n_samples // p)`` which creates non-overlapping test sets. Due to the high number of iterations which grows combinatorically with the number of samples this cross-validation method can be very costly. For large datasets one should favor :class:`KFold`, :class:`StratifiedKFold` or :class:`ShuffleSplit`. Read more in the :ref:`User Guide `. Parameters ---------- p : int Size of the test sets. Examples -------- >>> from sklearn.model_selection import LeavePOut >>> X = np.array([[1, 2], [3, 4], [5, 6], [7, 8]]) >>> y = np.array([1, 2, 3, 4]) >>> lpo = LeavePOut(2) >>> lpo.get_n_splits(X) 6 >>> print(lpo) LeavePOut(p=2) >>> for train_index, test_index in lpo.split(X): ... print("TRAIN:", train_index, "TEST:", test_index) ... X_train, X_test = X[train_index], X[test_index] ... y_train, y_test = y[train_index], y[test_index] TRAIN: [2 3] TEST: [0 1] TRAIN: [1 3] TEST: [0 2] TRAIN: [1 2] TEST: [0 3] TRAIN: [0 3] TEST: [1 2] TRAIN: [0 2] TEST: [1 3] TRAIN: [0 1] TEST: [2 3] """ def __init__(self, p): self.p = p def _iter_test_indices(self, X, y=None, groups=None): for combination in combinations(range(_num_samples(X)), self.p): yield np.array(combination) def get_n_splits(self, X, y=None, groups=None): """Returns the number of splitting iterations in the cross-validator Parameters ---------- X : array-like, shape (n_samples, n_features) Training data, where n_samples is the number of samples and n_features is the number of features. y : object Always ignored, exists for compatibility. groups : object Always ignored, exists for compatibility. """ if X is None: raise ValueError("The 'X' parameter should not be None.") return int(comb(_num_samples(X), self.p, exact=True)) class _BaseKFold(with_metaclass(ABCMeta, BaseCrossValidator)): """Base class for KFold, GroupKFold, and StratifiedKFold""" @abstractmethod def __init__(self, n_splits, shuffle, random_state): if not isinstance(n_splits, numbers.Integral): raise ValueError('The number of folds must be of Integral type. ' '%s of type %s was passed.' % (n_splits, type(n_splits))) n_splits = int(n_splits) if n_splits <= 1: raise ValueError( "k-fold cross-validation requires at least one" " train/test split by setting n_splits=2 or more," " got n_splits={0}.".format(n_splits)) if not isinstance(shuffle, bool): raise TypeError("shuffle must be True or False;" " got {0}".format(shuffle)) self.n_splits = n_splits self.shuffle = shuffle self.random_state = random_state def split(self, X, y=None, groups=None): """Generate indices to split data into training and test set. Parameters ---------- X : array-like, shape (n_samples, n_features) Training data, where n_samples is the number of samples and n_features is the number of features. y : array-like, shape (n_samples,) The target variable for supervised learning problems. groups : array-like, with shape (n_samples,), optional Group labels for the samples used while splitting the dataset into train/test set. Returns ------- train : ndarray The training set indices for that split. test : ndarray The testing set indices for that split. Notes ----- Randomized CV splitters may return different results for each call of split. You can make the results identical by setting ``random_state`` to an integer. """ X, y, groups = indexable(X, y, groups) n_samples = _num_samples(X) if self.n_splits > n_samples: raise ValueError( ("Cannot have number of splits n_splits={0} greater" " than the number of samples: {1}.").format(self.n_splits, n_samples)) for train, test in super(_BaseKFold, self).split(X, y, groups): yield train, test def get_n_splits(self, X=None, y=None, groups=None): """Returns the number of splitting iterations in the cross-validator Parameters ---------- X : object Always ignored, exists for compatibility. y : object Always ignored, exists for compatibility. groups : object Always ignored, exists for compatibility. Returns ------- n_splits : int Returns the number of splitting iterations in the cross-validator. """ return self.n_splits class KFold(_BaseKFold): """K-Folds cross-validator Provides train/test indices to split data in train/test sets. Split dataset into k consecutive folds (without shuffling by default). Each fold is then used once as a validation while the k - 1 remaining folds form the training set. Read more in the :ref:`User Guide `. Parameters ---------- n_splits : int, default=3 Number of folds. Must be at least 2. shuffle : boolean, optional Whether to shuffle the data before splitting into batches. random_state : int, RandomState instance or None, optional, default=None If int, random_state is the seed used by the random number generator; If RandomState instance, random_state is the random number generator; If None, the random number generator is the RandomState instance used by `np.random`. Used when ``shuffle`` == True. Examples -------- >>> from sklearn.model_selection import KFold >>> X = np.array([[1, 2], [3, 4], [1, 2], [3, 4]]) >>> y = np.array([1, 2, 3, 4]) >>> kf = KFold(n_splits=2) >>> kf.get_n_splits(X) 2 >>> print(kf) # doctest: +NORMALIZE_WHITESPACE KFold(n_splits=2, random_state=None, shuffle=False) >>> for train_index, test_index in kf.split(X): ... print("TRAIN:", train_index, "TEST:", test_index) ... X_train, X_test = X[train_index], X[test_index] ... y_train, y_test = y[train_index], y[test_index] TRAIN: [2 3] TEST: [0 1] TRAIN: [0 1] TEST: [2 3] Notes ----- The first ``n_samples % n_splits`` folds have size ``n_samples // n_splits + 1``, other folds have size ``n_samples // n_splits``, where ``n_samples`` is the number of samples. See also -------- StratifiedKFold Takes group information into account to avoid building folds with imbalanced class distributions (for binary or multiclass classification tasks). GroupKFold: K-fold iterator variant with non-overlapping groups. RepeatedKFold: Repeats K-Fold n times. """ def __init__(self, n_splits=3, shuffle=False, random_state=None): super(KFold, self).__init__(n_splits, shuffle, random_state) def _iter_test_indices(self, X, y=None, groups=None): n_samples = _num_samples(X) indices = np.arange(n_samples) if self.shuffle: check_random_state(self.random_state).shuffle(indices) n_splits = self.n_splits fold_sizes = (n_samples // n_splits) * np.ones(n_splits, dtype=np.int) fold_sizes[:n_samples % n_splits] += 1 current = 0 for fold_size in fold_sizes: start, stop = current, current + fold_size yield indices[start:stop] current = stop class GroupKFold(_BaseKFold): """K-fold iterator variant with non-overlapping groups. The same group will not appear in two different folds (the number of distinct groups has to be at least equal to the number of folds). The folds are approximately balanced in the sense that the number of distinct groups is approximately the same in each fold. Parameters ---------- n_splits : int, default=3 Number of folds. Must be at least 2. Examples -------- >>> from sklearn.model_selection import GroupKFold >>> X = np.array([[1, 2], [3, 4], [5, 6], [7, 8]]) >>> y = np.array([1, 2, 3, 4]) >>> groups = np.array([0, 0, 2, 2]) >>> group_kfold = GroupKFold(n_splits=2) >>> group_kfold.get_n_splits(X, y, groups) 2 >>> print(group_kfold) GroupKFold(n_splits=2) >>> for train_index, test_index in group_kfold.split(X, y, groups): ... print("TRAIN:", train_index, "TEST:", test_index) ... X_train, X_test = X[train_index], X[test_index] ... y_train, y_test = y[train_index], y[test_index] ... print(X_train, X_test, y_train, y_test) ... TRAIN: [0 1] TEST: [2 3] [[1 2] [3 4]] [[5 6] [7 8]] [1 2] [3 4] TRAIN: [2 3] TEST: [0 1] [[5 6] [7 8]] [[1 2] [3 4]] [3 4] [1 2] See also -------- LeaveOneGroupOut For splitting the data according to explicit domain-specific stratification of the dataset. """ def __init__(self, n_splits=3): super(GroupKFold, self).__init__(n_splits, shuffle=False, random_state=None) def _iter_test_indices(self, X, y, groups): if groups is None: raise ValueError("The 'groups' parameter should not be None.") groups = check_array(groups, ensure_2d=False, dtype=None) unique_groups, groups = np.unique(groups, return_inverse=True) n_groups = len(unique_groups) if self.n_splits > n_groups: raise ValueError("Cannot have number of splits n_splits=%d greater" " than the number of groups: %d." % (self.n_splits, n_groups)) # Weight groups by their number of occurrences n_samples_per_group = np.bincount(groups) # Distribute the most frequent groups first indices = np.argsort(n_samples_per_group)[::-1] n_samples_per_group = n_samples_per_group[indices] # Total weight of each fold n_samples_per_fold = np.zeros(self.n_splits) # Mapping from group index to fold index group_to_fold = np.zeros(len(unique_groups)) # Distribute samples by adding the largest weight to the lightest fold for group_index, weight in enumerate(n_samples_per_group): lightest_fold = np.argmin(n_samples_per_fold) n_samples_per_fold[lightest_fold] += weight group_to_fold[indices[group_index]] = lightest_fold indices = group_to_fold[groups] for f in range(self.n_splits): yield np.where(indices == f)[0] class StratifiedKFold(_BaseKFold): """Stratified K-Folds cross-validator Provides train/test indices to split data in train/test sets. This cross-validation object is a variation of KFold that returns stratified folds. The folds are made by preserving the percentage of samples for each class. Read more in the :ref:`User Guide `. Parameters ---------- n_splits : int, default=3 Number of folds. Must be at least 2. shuffle : boolean, optional Whether to shuffle each stratification of the data before splitting into batches. random_state : int, RandomState instance or None, optional, default=None If int, random_state is the seed used by the random number generator; If RandomState instance, random_state is the random number generator; If None, the random number generator is the RandomState instance used by `np.random`. Used when ``shuffle`` == True. Examples -------- >>> from sklearn.model_selection import StratifiedKFold >>> X = np.array([[1, 2], [3, 4], [1, 2], [3, 4]]) >>> y = np.array([0, 0, 1, 1]) >>> skf = StratifiedKFold(n_splits=2) >>> skf.get_n_splits(X, y) 2 >>> print(skf) # doctest: +NORMALIZE_WHITESPACE StratifiedKFold(n_splits=2, random_state=None, shuffle=False) >>> for train_index, test_index in skf.split(X, y): ... print("TRAIN:", train_index, "TEST:", test_index) ... X_train, X_test = X[train_index], X[test_index] ... y_train, y_test = y[train_index], y[test_index] TRAIN: [1 3] TEST: [0 2] TRAIN: [0 2] TEST: [1 3] Notes ----- All the folds have size ``trunc(n_samples / n_splits)``, the last one has the complementary. See also -------- RepeatedStratifiedKFold: Repeats Stratified K-Fold n times. """ def __init__(self, n_splits=3, shuffle=False, random_state=None): super(StratifiedKFold, self).__init__(n_splits, shuffle, random_state) def _make_test_folds(self, X, y=None): rng = self.random_state y = np.asarray(y) type_of_target_y = type_of_target(y) allowed_target_types = ('binary', 'multiclass') if type_of_target_y not in allowed_target_types: raise ValueError( 'Supported target types are: {}. Got {!r} instead.'.format( allowed_target_types, type_of_target_y)) y = column_or_1d(y) n_samples = y.shape[0] unique_y, y_inversed = np.unique(y, return_inverse=True) y_counts = np.bincount(y_inversed) min_groups = np.min(y_counts) if np.all(self.n_splits > y_counts): raise ValueError("n_splits=%d cannot be greater than the" " number of members in each class." % (self.n_splits)) if self.n_splits > min_groups: warnings.warn(("The least populated class in y has only %d" " members, which is too few. The minimum" " number of members in any class cannot" " be less than n_splits=%d." % (min_groups, self.n_splits)), Warning) # pre-assign each sample to a test fold index using individual KFold # splitting strategies for each class so as to respect the balance of # classes # NOTE: Passing the data corresponding to ith class say X[y==class_i] # will break when the data is not 100% stratifiable for all classes. # So we pass np.zeroes(max(c, n_splits)) as data to the KFold per_cls_cvs = [ KFold(self.n_splits, shuffle=self.shuffle, random_state=rng).split(np.zeros(max(count, self.n_splits))) for count in y_counts] test_folds = np.zeros(n_samples, dtype=np.int) for test_fold_indices, per_cls_splits in enumerate(zip(*per_cls_cvs)): for cls, (_, test_split) in zip(unique_y, per_cls_splits): cls_test_folds = test_folds[y == cls] # the test split can be too big because we used # KFold(...).split(X[:max(c, n_splits)]) when data is not 100% # stratifiable for all the classes # (we use a warning instead of raising an exception) # If this is the case, let's trim it: test_split = test_split[test_split < len(cls_test_folds)] cls_test_folds[test_split] = test_fold_indices test_folds[y == cls] = cls_test_folds return test_folds def _iter_test_masks(self, X, y=None, groups=None): test_folds = self._make_test_folds(X, y) for i in range(self.n_splits): yield test_folds == i def split(self, X, y, groups=None): """Generate indices to split data into training and test set. Parameters ---------- X : array-like, shape (n_samples, n_features) Training data, where n_samples is the number of samples and n_features is the number of features. Note that providing ``y`` is sufficient to generate the splits and hence ``np.zeros(n_samples)`` may be used as a placeholder for ``X`` instead of actual training data. y : array-like, shape (n_samples,) The target variable for supervised learning problems. Stratification is done based on the y labels. groups : object Always ignored, exists for compatibility. Returns ------- train : ndarray The training set indices for that split. test : ndarray The testing set indices for that split. Notes ----- Randomized CV splitters may return different results for each call of split. You can make the results identical by setting ``random_state`` to an integer. """ y = check_array(y, ensure_2d=False, dtype=None) return super(StratifiedKFold, self).split(X, y, groups) class TimeSeriesSplit(_BaseKFold): """Time Series cross-validator Provides train/test indices to split time series data samples that are observed at fixed time intervals, in train/test sets. In each split, test indices must be higher than before, and thus shuffling in cross validator is inappropriate. This cross-validation object is a variation of :class:`KFold`. In the kth split, it returns first k folds as train set and the (k+1)th fold as test set. Note that unlike standard cross-validation methods, successive training sets are supersets of those that come before them. Read more in the :ref:`User Guide `. Parameters ---------- n_splits : int, default=3 Number of splits. Must be at least 1. max_train_size : int, optional Maximum size for a single training set. Examples -------- >>> from sklearn.model_selection import TimeSeriesSplit >>> X = np.array([[1, 2], [3, 4], [1, 2], [3, 4]]) >>> y = np.array([1, 2, 3, 4]) >>> tscv = TimeSeriesSplit(n_splits=3) >>> print(tscv) # doctest: +NORMALIZE_WHITESPACE TimeSeriesSplit(max_train_size=None, n_splits=3) >>> for train_index, test_index in tscv.split(X): ... print("TRAIN:", train_index, "TEST:", test_index) ... X_train, X_test = X[train_index], X[test_index] ... y_train, y_test = y[train_index], y[test_index] TRAIN: [0] TEST: [1] TRAIN: [0 1] TEST: [2] TRAIN: [0 1 2] TEST: [3] Notes ----- The training set has size ``i * n_samples // (n_splits + 1) + n_samples % (n_splits + 1)`` in the ``i``th split, with a test set of size ``n_samples//(n_splits + 1)``, where ``n_samples`` is the number of samples. """ def __init__(self, n_splits=3, max_train_size=None): super(TimeSeriesSplit, self).__init__(n_splits, shuffle=False, random_state=None) self.max_train_size = max_train_size def split(self, X, y=None, groups=None): """Generate indices to split data into training and test set. Parameters ---------- X : array-like, shape (n_samples, n_features) Training data, where n_samples is the number of samples and n_features is the number of features. y : array-like, shape (n_samples,) Always ignored, exists for compatibility. groups : array-like, with shape (n_samples,), optional Always ignored, exists for compatibility. Returns ------- train : ndarray The training set indices for that split. test : ndarray The testing set indices for that split. Notes ----- Randomized CV splitters may return different results for each call of split. You can make the results identical by setting ``random_state`` to an integer. """ X, y, groups = indexable(X, y, groups) n_samples = _num_samples(X) n_splits = self.n_splits n_folds = n_splits + 1 if n_folds > n_samples: raise ValueError( ("Cannot have number of folds ={0} greater" " than the number of samples: {1}.").format(n_folds, n_samples)) indices = np.arange(n_samples) test_size = (n_samples // n_folds) test_starts = range(test_size + n_samples % n_folds, n_samples, test_size) for test_start in test_starts: if self.max_train_size and self.max_train_size < test_start: yield (indices[test_start - self.max_train_size:test_start], indices[test_start:test_start + test_size]) else: yield (indices[:test_start], indices[test_start:test_start + test_size]) class LeaveOneGroupOut(BaseCrossValidator): """Leave One Group Out cross-validator Provides train/test indices to split data according to a third-party provided group. This group information can be used to encode arbitrary domain specific stratifications of the samples as integers. For instance the groups could be the year of collection of the samples and thus allow for cross-validation against time-based splits. Read more in the :ref:`User Guide `. Examples -------- >>> from sklearn.model_selection import LeaveOneGroupOut >>> X = np.array([[1, 2], [3, 4], [5, 6], [7, 8]]) >>> y = np.array([1, 2, 1, 2]) >>> groups = np.array([1, 1, 2, 2]) >>> logo = LeaveOneGroupOut() >>> logo.get_n_splits(X, y, groups) 2 >>> logo.get_n_splits(groups=groups) # 'groups' is always required 2 >>> print(logo) LeaveOneGroupOut() >>> for train_index, test_index in logo.split(X, y, groups): ... print("TRAIN:", train_index, "TEST:", test_index) ... X_train, X_test = X[train_index], X[test_index] ... y_train, y_test = y[train_index], y[test_index] ... print(X_train, X_test, y_train, y_test) TRAIN: [2 3] TEST: [0 1] [[5 6] [7 8]] [[1 2] [3 4]] [1 2] [1 2] TRAIN: [0 1] TEST: [2 3] [[1 2] [3 4]] [[5 6] [7 8]] [1 2] [1 2] """ def _iter_test_masks(self, X, y, groups): if groups is None: raise ValueError("The 'groups' parameter should not be None.") # We make a copy of groups to avoid side-effects during iteration groups = check_array(groups, copy=True, ensure_2d=False, dtype=None) unique_groups = np.unique(groups) if len(unique_groups) <= 1: raise ValueError( "The groups parameter contains fewer than 2 unique groups " "(%s). LeaveOneGroupOut expects at least 2." % unique_groups) for i in unique_groups: yield groups == i def get_n_splits(self, X=None, y=None, groups=None): """Returns the number of splitting iterations in the cross-validator Parameters ---------- X : object, optional Always ignored, exists for compatibility. y : object, optional Always ignored, exists for compatibility. groups : array-like, with shape (n_samples,), optional Group labels for the samples used while splitting the dataset into train/test set. This 'groups' parameter must always be specified to calculate the number of splits, though the other parameters can be omitted. Returns ------- n_splits : int Returns the number of splitting iterations in the cross-validator. """ if groups is None: raise ValueError("The 'groups' parameter should not be None.") groups = check_array(groups, ensure_2d=False, dtype=None) return len(np.unique(groups)) class LeavePGroupsOut(BaseCrossValidator): """Leave P Group(s) Out cross-validator Provides train/test indices to split data according to a third-party provided group. This group information can be used to encode arbitrary domain specific stratifications of the samples as integers. For instance the groups could be the year of collection of the samples and thus allow for cross-validation against time-based splits. The difference between LeavePGroupsOut and LeaveOneGroupOut is that the former builds the test sets with all the samples assigned to ``p`` different values of the groups while the latter uses samples all assigned the same groups. Read more in the :ref:`User Guide `. Parameters ---------- n_groups : int Number of groups (``p``) to leave out in the test split. Examples -------- >>> from sklearn.model_selection import LeavePGroupsOut >>> X = np.array([[1, 2], [3, 4], [5, 6]]) >>> y = np.array([1, 2, 1]) >>> groups = np.array([1, 2, 3]) >>> lpgo = LeavePGroupsOut(n_groups=2) >>> lpgo.get_n_splits(X, y, groups) 3 >>> lpgo.get_n_splits(groups=groups) # 'groups' is always required 3 >>> print(lpgo) LeavePGroupsOut(n_groups=2) >>> for train_index, test_index in lpgo.split(X, y, groups): ... print("TRAIN:", train_index, "TEST:", test_index) ... X_train, X_test = X[train_index], X[test_index] ... y_train, y_test = y[train_index], y[test_index] ... print(X_train, X_test, y_train, y_test) TRAIN: [2] TEST: [0 1] [[5 6]] [[1 2] [3 4]] [1] [1 2] TRAIN: [1] TEST: [0 2] [[3 4]] [[1 2] [5 6]] [2] [1 1] TRAIN: [0] TEST: [1 2] [[1 2]] [[3 4] [5 6]] [1] [2 1] See also -------- GroupKFold: K-fold iterator variant with non-overlapping groups. """ def __init__(self, n_groups): self.n_groups = n_groups def _iter_test_masks(self, X, y, groups): if groups is None: raise ValueError("The 'groups' parameter should not be None.") groups = check_array(groups, copy=True, ensure_2d=False, dtype=None) unique_groups = np.unique(groups) if self.n_groups >= len(unique_groups): raise ValueError( "The groups parameter contains fewer than (or equal to) " "n_groups (%d) numbers of unique groups (%s). LeavePGroupsOut " "expects that at least n_groups + 1 (%d) unique groups be " "present" % (self.n_groups, unique_groups, self.n_groups + 1)) combi = combinations(range(len(unique_groups)), self.n_groups) for indices in combi: test_index = np.zeros(_num_samples(X), dtype=np.bool) for l in unique_groups[np.array(indices)]: test_index[groups == l] = True yield test_index def get_n_splits(self, X=None, y=None, groups=None): """Returns the number of splitting iterations in the cross-validator Parameters ---------- X : object, optional Always ignored, exists for compatibility. y : object, optional Always ignored, exists for compatibility. groups : array-like, with shape (n_samples,), optional Group labels for the samples used while splitting the dataset into train/test set. This 'groups' parameter must always be specified to calculate the number of splits, though the other parameters can be omitted. Returns ------- n_splits : int Returns the number of splitting iterations in the cross-validator. """ if groups is None: raise ValueError("The 'groups' parameter should not be None.") groups = check_array(groups, ensure_2d=False, dtype=None) return int(comb(len(np.unique(groups)), self.n_groups, exact=True)) class _RepeatedSplits(with_metaclass(ABCMeta)): """Repeated splits for an arbitrary randomized CV splitter. Repeats splits for cross-validators n times with different randomization in each repetition. Parameters ---------- cv : callable Cross-validator class. n_repeats : int, default=10 Number of times cross-validator needs to be repeated. random_state : int, RandomState instance or None, optional, default=None If int, random_state is the seed used by the random number generator; If RandomState instance, random_state is the random number generator; If None, the random number generator is the RandomState instance used by `np.random`. **cvargs : additional params Constructor parameters for cv. Must not contain random_state and shuffle. """ def __init__(self, cv, n_repeats=10, random_state=None, **cvargs): if not isinstance(n_repeats, (np.integer, numbers.Integral)): raise ValueError("Number of repetitions must be of Integral type.") if n_repeats <= 0: raise ValueError("Number of repetitions must be greater than 0.") if any(key in cvargs for key in ('random_state', 'shuffle')): raise ValueError( "cvargs must not contain random_state or shuffle.") self.cv = cv self.n_repeats = n_repeats self.random_state = random_state self.cvargs = cvargs def split(self, X, y=None, groups=None): """Generates indices to split data into training and test set. Parameters ---------- X : array-like, shape (n_samples, n_features) Training data, where n_samples is the number of samples and n_features is the number of features. y : array-like, of length n_samples The target variable for supervised learning problems. groups : array-like, with shape (n_samples,), optional Group labels for the samples used while splitting the dataset into train/test set. Returns ------- train : ndarray The training set indices for that split. test : ndarray The testing set indices for that split. """ n_repeats = self.n_repeats rng = check_random_state(self.random_state) for idx in range(n_repeats): cv = self.cv(random_state=rng, shuffle=True, **self.cvargs) for train_index, test_index in cv.split(X, y, groups): yield train_index, test_index def get_n_splits(self, X=None, y=None, groups=None): """Returns the number of splitting iterations in the cross-validator Parameters ---------- X : object Always ignored, exists for compatibility. ``np.zeros(n_samples)`` may be used as a placeholder. y : object Always ignored, exists for compatibility. ``np.zeros(n_samples)`` may be used as a placeholder. groups : array-like, with shape (n_samples,), optional Group labels for the samples used while splitting the dataset into train/test set. Returns ------- n_splits : int Returns the number of splitting iterations in the cross-validator. """ rng = check_random_state(self.random_state) cv = self.cv(random_state=rng, shuffle=True, **self.cvargs) return cv.get_n_splits(X, y, groups) * self.n_repeats class RepeatedKFold(_RepeatedSplits): """Repeated K-Fold cross validator. Repeats K-Fold n times with different randomization in each repetition. Read more in the :ref:`User Guide `. Parameters ---------- n_splits : int, default=5 Number of folds. Must be at least 2. n_repeats : int, default=10 Number of times cross-validator needs to be repeated. random_state : int, RandomState instance or None, optional, default=None If int, random_state is the seed used by the random number generator; If RandomState instance, random_state is the random number generator; If None, the random number generator is the RandomState instance used by `np.random`. Examples -------- >>> from sklearn.model_selection import RepeatedKFold >>> X = np.array([[1, 2], [3, 4], [1, 2], [3, 4]]) >>> y = np.array([0, 0, 1, 1]) >>> rkf = RepeatedKFold(n_splits=2, n_repeats=2, random_state=2652124) >>> for train_index, test_index in rkf.split(X): ... print("TRAIN:", train_index, "TEST:", test_index) ... X_train, X_test = X[train_index], X[test_index] ... y_train, y_test = y[train_index], y[test_index] ... TRAIN: [0 1] TEST: [2 3] TRAIN: [2 3] TEST: [0 1] TRAIN: [1 2] TEST: [0 3] TRAIN: [0 3] TEST: [1 2] See also -------- RepeatedStratifiedKFold: Repeates Stratified K-Fold n times. """ def __init__(self, n_splits=5, n_repeats=10, random_state=None): super(RepeatedKFold, self).__init__( KFold, n_repeats, random_state, n_splits=n_splits) class RepeatedStratifiedKFold(_RepeatedSplits): """Repeated Stratified K-Fold cross validator. Repeats Stratified K-Fold n times with different randomization in each repetition. Read more in the :ref:`User Guide `. Parameters ---------- n_splits : int, default=5 Number of folds. Must be at least 2. n_repeats : int, default=10 Number of times cross-validator needs to be repeated. random_state : None, int or RandomState, default=None Random state to be used to generate random state for each repetition. Examples -------- >>> from sklearn.model_selection import RepeatedStratifiedKFold >>> X = np.array([[1, 2], [3, 4], [1, 2], [3, 4]]) >>> y = np.array([0, 0, 1, 1]) >>> rskf = RepeatedStratifiedKFold(n_splits=2, n_repeats=2, ... random_state=36851234) >>> for train_index, test_index in rskf.split(X, y): ... print("TRAIN:", train_index, "TEST:", test_index) ... X_train, X_test = X[train_index], X[test_index] ... y_train, y_test = y[train_index], y[test_index] ... TRAIN: [1 2] TEST: [0 3] TRAIN: [0 3] TEST: [1 2] TRAIN: [1 3] TEST: [0 2] TRAIN: [0 2] TEST: [1 3] See also -------- RepeatedKFold: Repeats K-Fold n times. """ def __init__(self, n_splits=5, n_repeats=10, random_state=None): super(RepeatedStratifiedKFold, self).__init__( StratifiedKFold, n_repeats, random_state, n_splits=n_splits) class BaseShuffleSplit(with_metaclass(ABCMeta)): """Base class for ShuffleSplit and StratifiedShuffleSplit""" def __init__(self, n_splits=10, test_size="default", train_size=None, random_state=None): _validate_shuffle_split_init(test_size, train_size) self.n_splits = n_splits self.test_size = test_size self.train_size = train_size self.random_state = random_state def split(self, X, y=None, groups=None): """Generate indices to split data into training and test set. Parameters ---------- X : array-like, shape (n_samples, n_features) Training data, where n_samples is the number of samples and n_features is the number of features. y : array-like, shape (n_samples,) The target variable for supervised learning problems. groups : array-like, with shape (n_samples,), optional Group labels for the samples used while splitting the dataset into train/test set. Returns ------- train : ndarray The training set indices for that split. test : ndarray The testing set indices for that split. Notes ----- Randomized CV splitters may return different results for each call of split. You can make the results identical by setting ``random_state`` to an integer. """ X, y, groups = indexable(X, y, groups) for train, test in self._iter_indices(X, y, groups): yield train, test @abstractmethod def _iter_indices(self, X, y=None, groups=None): """Generate (train, test) indices""" def get_n_splits(self, X=None, y=None, groups=None): """Returns the number of splitting iterations in the cross-validator Parameters ---------- X : object Always ignored, exists for compatibility. y : object Always ignored, exists for compatibility. groups : object Always ignored, exists for compatibility. Returns ------- n_splits : int Returns the number of splitting iterations in the cross-validator. """ return self.n_splits def __repr__(self): return _build_repr(self) class ShuffleSplit(BaseShuffleSplit): """Random permutation cross-validator Yields indices to split data into training and test sets. Note: contrary to other cross-validation strategies, random splits do not guarantee that all folds will be different, although this is still very likely for sizeable datasets. Read more in the :ref:`User Guide `. Parameters ---------- n_splits : int, default 10 Number of re-shuffling & splitting iterations. test_size : float, int, None, default=0.1 If float, should be between 0.0 and 1.0 and represent the proportion of the dataset to include in the test split. If int, represents the absolute number of test samples. If None, the value is set to the complement of the train size. By default (the is parameter unspecified), the value is set to 0.1. The default will change in version 0.21. It will remain 0.1 only if ``train_size`` is unspecified, otherwise it will complement the specified ``train_size``. train_size : float, int, or None, default=None If float, should be between 0.0 and 1.0 and represent the proportion of the dataset to include in the train split. If int, represents the absolute number of train samples. If None, the value is automatically set to the complement of the test size. random_state : int, RandomState instance or None, optional (default=None) If int, random_state is the seed used by the random number generator; If RandomState instance, random_state is the random number generator; If None, the random number generator is the RandomState instance used by `np.random`. Examples -------- >>> from sklearn.model_selection import ShuffleSplit >>> X = np.array([[1, 2], [3, 4], [5, 6], [7, 8]]) >>> y = np.array([1, 2, 1, 2]) >>> rs = ShuffleSplit(n_splits=3, test_size=.25, random_state=0) >>> rs.get_n_splits(X) 3 >>> print(rs) ShuffleSplit(n_splits=3, random_state=0, test_size=0.25, train_size=None) >>> for train_index, test_index in rs.split(X): ... print("TRAIN:", train_index, "TEST:", test_index) ... # doctest: +ELLIPSIS TRAIN: [3 1 0] TEST: [2] TRAIN: [2 1 3] TEST: [0] TRAIN: [0 2 1] TEST: [3] >>> rs = ShuffleSplit(n_splits=3, train_size=0.5, test_size=.25, ... random_state=0) >>> for train_index, test_index in rs.split(X): ... print("TRAIN:", train_index, "TEST:", test_index) ... # doctest: +ELLIPSIS TRAIN: [3 1] TEST: [2] TRAIN: [2 1] TEST: [0] TRAIN: [0 2] TEST: [3] """ def _iter_indices(self, X, y=None, groups=None): n_samples = _num_samples(X) n_train, n_test = _validate_shuffle_split(n_samples, self.test_size, self.train_size) rng = check_random_state(self.random_state) for i in range(self.n_splits): # random partition permutation = rng.permutation(n_samples) ind_test = permutation[:n_test] ind_train = permutation[n_test:(n_test + n_train)] yield ind_train, ind_test class GroupShuffleSplit(ShuffleSplit): '''Shuffle-Group(s)-Out cross-validation iterator Provides randomized train/test indices to split data according to a third-party provided group. This group information can be used to encode arbitrary domain specific stratifications of the samples as integers. For instance the groups could be the year of collection of the samples and thus allow for cross-validation against time-based splits. The difference between LeavePGroupsOut and GroupShuffleSplit is that the former generates splits using all subsets of size ``p`` unique groups, whereas GroupShuffleSplit generates a user-determined number of random test splits, each with a user-determined fraction of unique groups. For example, a less computationally intensive alternative to ``LeavePGroupsOut(p=10)`` would be ``GroupShuffleSplit(test_size=10, n_splits=100)``. Note: The parameters ``test_size`` and ``train_size`` refer to groups, and not to samples, as in ShuffleSplit. Parameters ---------- n_splits : int (default 5) Number of re-shuffling & splitting iterations. test_size : float, int, None, optional If float, should be between 0.0 and 1.0 and represent the proportion of the dataset to include in the test split. If int, represents the absolute number of test samples. If None, the value is set to the complement of the train size. By default, the value is set to 0.2. The default will change in version 0.21. It will remain 0.2 only if ``train_size`` is unspecified, otherwise it will complement the specified ``train_size``. train_size : float, int, or None, default is None If float, should be between 0.0 and 1.0 and represent the proportion of the groups to include in the train split. If int, represents the absolute number of train groups. If None, the value is automatically set to the complement of the test size. random_state : int, RandomState instance or None, optional (default=None) If int, random_state is the seed used by the random number generator; If RandomState instance, random_state is the random number generator; If None, the random number generator is the RandomState instance used by `np.random`. ''' def __init__(self, n_splits=5, test_size="default", train_size=None, random_state=None): if test_size == "default": if train_size is not None: warnings.warn("From version 0.21, test_size will always " "complement train_size unless both " "are specified.", FutureWarning) test_size = 0.2 super(GroupShuffleSplit, self).__init__( n_splits=n_splits, test_size=test_size, train_size=train_size, random_state=random_state) def _iter_indices(self, X, y, groups): if groups is None: raise ValueError("The 'groups' parameter should not be None.") groups = check_array(groups, ensure_2d=False, dtype=None) classes, group_indices = np.unique(groups, return_inverse=True) for group_train, group_test in super( GroupShuffleSplit, self)._iter_indices(X=classes): # these are the indices of classes in the partition # invert them into data indices train = np.flatnonzero(np.in1d(group_indices, group_train)) test = np.flatnonzero(np.in1d(group_indices, group_test)) yield train, test def _approximate_mode(class_counts, n_draws, rng): """Computes approximate mode of multivariate hypergeometric. This is an approximation to the mode of the multivariate hypergeometric given by class_counts and n_draws. It shouldn't be off by more than one. It is the mostly likely outcome of drawing n_draws many samples from the population given by class_counts. Parameters ---------- class_counts : ndarray of int Population per class. n_draws : int Number of draws (samples to draw) from the overall population. rng : random state Used to break ties. Returns ------- sampled_classes : ndarray of int Number of samples drawn from each class. np.sum(sampled_classes) == n_draws Examples -------- >>> from sklearn.model_selection._split import _approximate_mode >>> _approximate_mode(class_counts=np.array([4, 2]), n_draws=3, rng=0) array([2, 1]) >>> _approximate_mode(class_counts=np.array([5, 2]), n_draws=4, rng=0) array([3, 1]) >>> _approximate_mode(class_counts=np.array([2, 2, 2, 1]), ... n_draws=2, rng=0) array([0, 1, 1, 0]) >>> _approximate_mode(class_counts=np.array([2, 2, 2, 1]), ... n_draws=2, rng=42) array([1, 1, 0, 0]) """ rng = check_random_state(rng) # this computes a bad approximation to the mode of the # multivariate hypergeometric given by class_counts and n_draws continuous = n_draws * class_counts / class_counts.sum() # floored means we don't overshoot n_samples, but probably undershoot floored = np.floor(continuous) # we add samples according to how much "left over" probability # they had, until we arrive at n_samples need_to_add = int(n_draws - floored.sum()) if need_to_add > 0: remainder = continuous - floored values = np.sort(np.unique(remainder))[::-1] # add according to remainder, but break ties # randomly to avoid biases for value in values: inds, = np.where(remainder == value) # if we need_to_add less than what's in inds # we draw randomly from them. # if we need to add more, we add them all and # go to the next value add_now = min(len(inds), need_to_add) inds = rng.choice(inds, size=add_now, replace=False) floored[inds] += 1 need_to_add -= add_now if need_to_add == 0: break return floored.astype(np.int) class StratifiedShuffleSplit(BaseShuffleSplit): """Stratified ShuffleSplit cross-validator Provides train/test indices to split data in train/test sets. This cross-validation object is a merge of StratifiedKFold and ShuffleSplit, which returns stratified randomized folds. The folds are made by preserving the percentage of samples for each class. Note: like the ShuffleSplit strategy, stratified random splits do not guarantee that all folds will be different, although this is still very likely for sizeable datasets. Read more in the :ref:`User Guide `. Parameters ---------- n_splits : int, default 10 Number of re-shuffling & splitting iterations. test_size : float, int, None, optional If float, should be between 0.0 and 1.0 and represent the proportion of the dataset to include in the test split. If int, represents the absolute number of test samples. If None, the value is set to the complement of the train size. By default, the value is set to 0.1. The default will change in version 0.21. It will remain 0.1 only if ``train_size`` is unspecified, otherwise it will complement the specified ``train_size``. train_size : float, int, or None, default is None If float, should be between 0.0 and 1.0 and represent the proportion of the dataset to include in the train split. If int, represents the absolute number of train samples. If None, the value is automatically set to the complement of the test size. random_state : int, RandomState instance or None, optional (default=None) If int, random_state is the seed used by the random number generator; If RandomState instance, random_state is the random number generator; If None, the random number generator is the RandomState instance used by `np.random`. Examples -------- >>> from sklearn.model_selection import StratifiedShuffleSplit >>> X = np.array([[1, 2], [3, 4], [1, 2], [3, 4]]) >>> y = np.array([0, 0, 1, 1]) >>> sss = StratifiedShuffleSplit(n_splits=3, test_size=0.5, random_state=0) >>> sss.get_n_splits(X, y) 3 >>> print(sss) # doctest: +ELLIPSIS StratifiedShuffleSplit(n_splits=3, random_state=0, ...) >>> for train_index, test_index in sss.split(X, y): ... print("TRAIN:", train_index, "TEST:", test_index) ... X_train, X_test = X[train_index], X[test_index] ... y_train, y_test = y[train_index], y[test_index] TRAIN: [1 2] TEST: [3 0] TRAIN: [0 2] TEST: [1 3] TRAIN: [0 2] TEST: [3 1] """ def __init__(self, n_splits=10, test_size="default", train_size=None, random_state=None): super(StratifiedShuffleSplit, self).__init__( n_splits, test_size, train_size, random_state) def _iter_indices(self, X, y, groups=None): n_samples = _num_samples(X) y = check_array(y, ensure_2d=False, dtype=None) n_train, n_test = _validate_shuffle_split(n_samples, self.test_size, self.train_size) if y.ndim == 2: # for multi-label y, map each distinct row to a string repr # using join because str(row) uses an ellipsis if len(row) > 1000 y = np.array([' '.join(row.astype('str')) for row in y]) classes, y_indices = np.unique(y, return_inverse=True) n_classes = classes.shape[0] class_counts = np.bincount(y_indices) if np.min(class_counts) < 2: raise ValueError("The least populated class in y has only 1" " member, which is too few. The minimum" " number of groups for any class cannot" " be less than 2.") if n_train < n_classes: raise ValueError('The train_size = %d should be greater or ' 'equal to the number of classes = %d' % (n_train, n_classes)) if n_test < n_classes: raise ValueError('The test_size = %d should be greater or ' 'equal to the number of classes = %d' % (n_test, n_classes)) # Find the sorted list of instances for each class: # (np.unique above performs a sort, so code is O(n logn) already) class_indices = np.split(np.argsort(y_indices, kind='mergesort'), np.cumsum(class_counts)[:-1]) rng = check_random_state(self.random_state) for _ in range(self.n_splits): # if there are ties in the class-counts, we want # to make sure to break them anew in each iteration n_i = _approximate_mode(class_counts, n_train, rng) class_counts_remaining = class_counts - n_i t_i = _approximate_mode(class_counts_remaining, n_test, rng) train = [] test = [] for i in range(n_classes): permutation = rng.permutation(class_counts[i]) perm_indices_class_i = class_indices[i].take(permutation, mode='clip') train.extend(perm_indices_class_i[:n_i[i]]) test.extend(perm_indices_class_i[n_i[i]:n_i[i] + t_i[i]]) train = rng.permutation(train) test = rng.permutation(test) yield train, test def split(self, X, y, groups=None): """Generate indices to split data into training and test set. Parameters ---------- X : array-like, shape (n_samples, n_features) Training data, where n_samples is the number of samples and n_features is the number of features. Note that providing ``y`` is sufficient to generate the splits and hence ``np.zeros(n_samples)`` may be used as a placeholder for ``X`` instead of actual training data. y : array-like, shape (n_samples,) The target variable for supervised learning problems. Stratification is done based on the y labels. groups : object Always ignored, exists for compatibility. Returns ------- train : ndarray The training set indices for that split. test : ndarray The testing set indices for that split. Notes ----- Randomized CV splitters may return different results for each call of split. You can make the results identical by setting ``random_state`` to an integer. """ y = check_array(y, ensure_2d=False, dtype=None) return super(StratifiedShuffleSplit, self).split(X, y, groups) def _validate_shuffle_split_init(test_size, train_size): """Validation helper to check the test_size and train_size at init NOTE This does not take into account the number of samples which is known only at split """ if test_size == "default": if train_size is not None: warnings.warn("From version 0.21, test_size will always " "complement train_size unless both " "are specified.", FutureWarning) test_size = 0.1 if test_size is None and train_size is None: raise ValueError('test_size and train_size can not both be None') if test_size is not None: if np.asarray(test_size).dtype.kind == 'f': if test_size >= 1.: raise ValueError( 'test_size=%f should be smaller ' 'than 1.0 or be an integer' % test_size) elif np.asarray(test_size).dtype.kind != 'i': # int values are checked during split based on the input raise ValueError("Invalid value for test_size: %r" % test_size) if train_size is not None: if np.asarray(train_size).dtype.kind == 'f': if train_size >= 1.: raise ValueError("train_size=%f should be smaller " "than 1.0 or be an integer" % train_size) elif (np.asarray(test_size).dtype.kind == 'f' and (train_size + test_size) > 1.): raise ValueError('The sum of test_size and train_size = %f, ' 'should be smaller than 1.0. Reduce ' 'test_size and/or train_size.' % (train_size + test_size)) elif np.asarray(train_size).dtype.kind != 'i': # int values are checked during split based on the input raise ValueError("Invalid value for train_size: %r" % train_size) def _validate_shuffle_split(n_samples, test_size, train_size): """ Validation helper to check if the test/test sizes are meaningful wrt to the size of the data (n_samples) """ if (test_size is not None and np.asarray(test_size).dtype.kind == 'i' and test_size >= n_samples): raise ValueError('test_size=%d should be smaller than the number of ' 'samples %d' % (test_size, n_samples)) if (train_size is not None and np.asarray(train_size).dtype.kind == 'i' and train_size >= n_samples): raise ValueError("train_size=%d should be smaller than the number of" " samples %d" % (train_size, n_samples)) if test_size == "default": test_size = 0.1 if np.asarray(test_size).dtype.kind == 'f': n_test = ceil(test_size * n_samples) elif np.asarray(test_size).dtype.kind == 'i': n_test = float(test_size) if train_size is None: n_train = n_samples - n_test elif np.asarray(train_size).dtype.kind == 'f': n_train = floor(train_size * n_samples) else: n_train = float(train_size) if test_size is None: n_test = n_samples - n_train if n_train + n_test > n_samples: raise ValueError('The sum of train_size and test_size = %d, ' 'should be smaller than the number of ' 'samples %d. Reduce test_size and/or ' 'train_size.' % (n_train + n_test, n_samples)) return int(n_train), int(n_test) class PredefinedSplit(BaseCrossValidator): """Predefined split cross-validator Provides train/test indices to split data into train/test sets using a predefined scheme specified by the user with the ``test_fold`` parameter. Read more in the :ref:`User Guide `. Parameters ---------- test_fold : array-like, shape (n_samples,) The entry ``test_fold[i]`` represents the index of the test set that sample ``i`` belongs to. It is possible to exclude sample ``i`` from any test set (i.e. include sample ``i`` in every training set) by setting ``test_fold[i]`` equal to -1. Examples -------- >>> from sklearn.model_selection import PredefinedSplit >>> X = np.array([[1, 2], [3, 4], [1, 2], [3, 4]]) >>> y = np.array([0, 0, 1, 1]) >>> test_fold = [0, 1, -1, 1] >>> ps = PredefinedSplit(test_fold) >>> ps.get_n_splits() 2 >>> print(ps) # doctest: +NORMALIZE_WHITESPACE +ELLIPSIS PredefinedSplit(test_fold=array([ 0, 1, -1, 1])) >>> for train_index, test_index in ps.split(): ... print("TRAIN:", train_index, "TEST:", test_index) ... X_train, X_test = X[train_index], X[test_index] ... y_train, y_test = y[train_index], y[test_index] TRAIN: [1 2 3] TEST: [0] TRAIN: [0 2] TEST: [1 3] """ def __init__(self, test_fold): self.test_fold = np.array(test_fold, dtype=np.int) self.test_fold = column_or_1d(self.test_fold) self.unique_folds = np.unique(self.test_fold) self.unique_folds = self.unique_folds[self.unique_folds != -1] def split(self, X=None, y=None, groups=None): """Generate indices to split data into training and test set. Parameters ---------- X : object Always ignored, exists for compatibility. y : object Always ignored, exists for compatibility. groups : object Always ignored, exists for compatibility. Returns ------- train : ndarray The training set indices for that split. test : ndarray The testing set indices for that split. """ ind = np.arange(len(self.test_fold)) for test_index in self._iter_test_masks(): train_index = ind[np.logical_not(test_index)] test_index = ind[test_index] yield train_index, test_index def _iter_test_masks(self): """Generates boolean masks corresponding to test sets.""" for f in self.unique_folds: test_index = np.where(self.test_fold == f)[0] test_mask = np.zeros(len(self.test_fold), dtype=np.bool) test_mask[test_index] = True yield test_mask def get_n_splits(self, X=None, y=None, groups=None): """Returns the number of splitting iterations in the cross-validator Parameters ---------- X : object Always ignored, exists for compatibility. y : object Always ignored, exists for compatibility. groups : object Always ignored, exists for compatibility. Returns ------- n_splits : int Returns the number of splitting iterations in the cross-validator. """ return len(self.unique_folds) class _CVIterableWrapper(BaseCrossValidator): """Wrapper class for old style cv objects and iterables.""" def __init__(self, cv): self.cv = list(cv) def get_n_splits(self, X=None, y=None, groups=None): """Returns the number of splitting iterations in the cross-validator Parameters ---------- X : object Always ignored, exists for compatibility. y : object Always ignored, exists for compatibility. groups : object Always ignored, exists for compatibility. Returns ------- n_splits : int Returns the number of splitting iterations in the cross-validator. """ return len(self.cv) def split(self, X=None, y=None, groups=None): """Generate indices to split data into training and test set. Parameters ---------- X : object Always ignored, exists for compatibility. y : object Always ignored, exists for compatibility. groups : object Always ignored, exists for compatibility. Returns ------- train : ndarray The training set indices for that split. test : ndarray The testing set indices for that split. """ for train, test in self.cv: yield train, test def check_cv(cv=3, y=None, classifier=False): """Input checker utility for building a cross-validator Parameters ---------- cv : int, cross-validation generator or an iterable, optional Determines the cross-validation splitting strategy. Possible inputs for cv are: - None, to use the default 3-fold cross-validation, - integer, to specify the number of folds. - An object to be used as a cross-validation generator. - An iterable yielding train/test splits. For integer/None inputs, if classifier is True and ``y`` is either binary or multiclass, :class:`StratifiedKFold` is used. In all other cases, :class:`KFold` is used. Refer :ref:`User Guide ` for the various cross-validation strategies that can be used here. y : array-like, optional The target variable for supervised learning problems. classifier : boolean, optional, default False Whether the task is a classification task, in which case stratified KFold will be used. Returns ------- checked_cv : a cross-validator instance. The return value is a cross-validator which generates the train/test splits via the ``split`` method. """ if cv is None: cv = 3 if isinstance(cv, numbers.Integral): if (classifier and (y is not None) and (type_of_target(y) in ('binary', 'multiclass'))): return StratifiedKFold(cv) else: return KFold(cv) if not hasattr(cv, 'split') or isinstance(cv, str): if not isinstance(cv, Iterable) or isinstance(cv, str): raise ValueError("Expected cv as an integer, cross-validation " "object (from sklearn.model_selection) " "or an iterable. Got %s." % cv) return _CVIterableWrapper(cv) return cv # New style cv objects are passed without any modification def train_test_split(*arrays, **options): """Split arrays or matrices into random train and test subsets Quick utility that wraps input validation and ``next(ShuffleSplit().split(X, y))`` and application to input data into a single call for splitting (and optionally subsampling) data in a oneliner. Read more in the :ref:`User Guide `. Parameters ---------- *arrays : sequence of indexables with same length / shape[0] Allowed inputs are lists, numpy arrays, scipy-sparse matrices or pandas dataframes. test_size : float, int, None, optional If float, should be between 0.0 and 1.0 and represent the proportion of the dataset to include in the test split. If int, represents the absolute number of test samples. If None, the value is set to the complement of the train size. By default, the value is set to 0.25. The default will change in version 0.21. It will remain 0.25 only if ``train_size`` is unspecified, otherwise it will complement the specified ``train_size``. train_size : float, int, or None, default None If float, should be between 0.0 and 1.0 and represent the proportion of the dataset to include in the train split. If int, represents the absolute number of train samples. If None, the value is automatically set to the complement of the test size. random_state : int, RandomState instance or None, optional (default=None) If int, random_state is the seed used by the random number generator; If RandomState instance, random_state is the random number generator; If None, the random number generator is the RandomState instance used by `np.random`. shuffle : boolean, optional (default=True) Whether or not to shuffle the data before splitting. If shuffle=False then stratify must be None. stratify : array-like or None (default is None) If not None, data is split in a stratified fashion, using this as the class labels. Returns ------- splitting : list, length=2 * len(arrays) List containing train-test split of inputs. .. versionadded:: 0.16 If the input is sparse, the output will be a ``scipy.sparse.csr_matrix``. Else, output type is the same as the input type. Examples -------- >>> import numpy as np >>> from sklearn.model_selection import train_test_split >>> X, y = np.arange(10).reshape((5, 2)), range(5) >>> X array([[0, 1], [2, 3], [4, 5], [6, 7], [8, 9]]) >>> list(y) [0, 1, 2, 3, 4] >>> X_train, X_test, y_train, y_test = train_test_split( ... X, y, test_size=0.33, random_state=42) ... >>> X_train array([[4, 5], [0, 1], [6, 7]]) >>> y_train [2, 0, 3] >>> X_test array([[2, 3], [8, 9]]) >>> y_test [1, 4] >>> train_test_split(y, shuffle=False) [[0, 1, 2], [3, 4]] """ n_arrays = len(arrays) if n_arrays == 0: raise ValueError("At least one array required as input") test_size = options.pop('test_size', 'default') train_size = options.pop('train_size', None) random_state = options.pop('random_state', None) stratify = options.pop('stratify', None) shuffle = options.pop('shuffle', True) if options: raise TypeError("Invalid parameters passed: %s" % str(options)) if test_size == 'default': test_size = None if train_size is not None: warnings.warn("From version 0.21, test_size will always " "complement train_size unless both " "are specified.", FutureWarning) if test_size is None and train_size is None: test_size = 0.25 arrays = indexable(*arrays) if shuffle is False: if stratify is not None: raise ValueError( "Stratified train/test split is not implemented for " "shuffle=False") n_samples = _num_samples(arrays[0]) n_train, n_test = _validate_shuffle_split(n_samples, test_size, train_size) train = np.arange(n_train) test = np.arange(n_train, n_train + n_test) else: if stratify is not None: CVClass = StratifiedShuffleSplit else: CVClass = ShuffleSplit cv = CVClass(test_size=test_size, train_size=train_size, random_state=random_state) train, test = next(cv.split(X=arrays[0], y=stratify)) return list(chain.from_iterable((safe_indexing(a, train), safe_indexing(a, test)) for a in arrays)) train_test_split.__test__ = False # to avoid a pb with nosetests def _build_repr(self): # XXX This is copied from BaseEstimator's get_params cls = self.__class__ init = getattr(cls.__init__, 'deprecated_original', cls.__init__) # Ignore varargs, kw and default values and pop self init_signature = signature(init) # Consider the constructor parameters excluding 'self' if init is object.__init__: args = [] else: args = sorted([p.name for p in init_signature.parameters.values() if p.name != 'self' and p.kind != p.VAR_KEYWORD]) class_name = self.__class__.__name__ params = dict() for key in args: # We need deprecation warnings to always be on in order to # catch deprecated param values. # This is set in utils/__init__.py but it gets overwritten # when running under python3 somehow. warnings.simplefilter("always", DeprecationWarning) try: with warnings.catch_warnings(record=True) as w: value = getattr(self, key, None) if len(w) and w[0].category == DeprecationWarning: # if the parameter is deprecated, don't show it continue finally: warnings.filters.pop(0) params[key] = value return '%s(%s)' % (class_name, _pprint(params, offset=len(class_name)))