U 2dH=@sddlmZddlZddlZddlmZddlmZm Z ddl m Z m Z m Z mZddlmZdd lmZdd lmZdd lmZdd lmZGd dde eZdS))IntegralN) OneHotEncoder) BaseEstimatorTransformerMixin)HiddenInterval StrOptionsOptions) check_array)check_is_fitted)check_random_state)_check_feature_names_in)_safe_indexingc @seZdZUdZeedddddgeddd hged d d hgeee j e j hdgeed dddde edhgdgdZ eed<d dd ddddddZd!ddZddZddZddZd"ddZdS)#KBinsDiscretizera Bin continuous data into intervals. Read more in the :ref:`User Guide `. .. versionadded:: 0.20 Parameters ---------- n_bins : int or array-like of shape (n_features,), default=5 The number of bins to produce. Raises ValueError if ``n_bins < 2``. encode : {'onehot', 'onehot-dense', 'ordinal'}, default='onehot' Method used to encode the transformed result. - 'onehot': Encode the transformed result with one-hot encoding and return a sparse matrix. Ignored features are always stacked to the right. - 'onehot-dense': Encode the transformed result with one-hot encoding and return a dense array. Ignored features are always stacked to the right. - 'ordinal': Return the bin identifier encoded as an integer value. strategy : {'uniform', 'quantile', 'kmeans'}, default='quantile' Strategy used to define the widths of the bins. - 'uniform': All bins in each feature have identical widths. - 'quantile': All bins in each feature have the same number of points. - 'kmeans': Values in each bin have the same nearest center of a 1D k-means cluster. dtype : {np.float32, np.float64}, default=None The desired data-type for the output. If None, output dtype is consistent with input dtype. Only np.float32 and np.float64 are supported. .. versionadded:: 0.24 subsample : int or None (default='warn') Maximum number of samples, used to fit the model, for computational efficiency. Used when `strategy="quantile"`. `subsample=None` means that all the training samples are used when computing the quantiles that determine the binning thresholds. Since quantile computation relies on sorting each column of `X` and that sorting has an `n log(n)` time complexity, it is recommended to use subsampling on datasets with a very large number of samples. .. deprecated:: 1.1 In version 1.3 and onwards, `subsample=2e5` will be the default. random_state : int, RandomState instance or None, default=None Determines random number generation for subsampling. Pass an int for reproducible results across multiple function calls. See the `subsample` parameter for more details. See :term:`Glossary `. .. versionadded:: 1.1 Attributes ---------- bin_edges_ : ndarray of ndarray of shape (n_features,) The edges of each bin. Contain arrays of varying shapes ``(n_bins_, )`` Ignored features will have empty arrays. n_bins_ : ndarray of shape (n_features,), dtype=np.int_ Number of bins per feature. Bins whose width are too small (i.e., <= 1e-8) are removed with a warning. n_features_in_ : int Number of features seen during :term:`fit`. .. versionadded:: 0.24 feature_names_in_ : ndarray of shape (`n_features_in_`,) Names of features seen during :term:`fit`. Defined only when `X` has feature names that are all strings. .. versionadded:: 1.0 See Also -------- Binarizer : Class used to bin values as ``0`` or ``1`` based on a parameter ``threshold``. Notes ----- In bin edges for feature ``i``, the first and last values are used only for ``inverse_transform``. During transform, bin edges are extended to:: np.concatenate([-np.inf, bin_edges_[i][1:-1], np.inf]) You can combine ``KBinsDiscretizer`` with :class:`~sklearn.compose.ColumnTransformer` if you only want to preprocess part of the features. ``KBinsDiscretizer`` might produce constant features (e.g., when ``encode = 'onehot'`` and certain bins do not contain any data). These features can be removed with feature selection algorithms (e.g., :class:`~sklearn.feature_selection.VarianceThreshold`). Examples -------- >>> from sklearn.preprocessing import KBinsDiscretizer >>> X = [[-2, 1, -4, -1], ... [-1, 2, -3, -0.5], ... [ 0, 3, -2, 0.5], ... [ 1, 4, -1, 2]] >>> est = KBinsDiscretizer(n_bins=3, encode='ordinal', strategy='uniform') >>> est.fit(X) KBinsDiscretizer(...) >>> Xt = est.transform(X) >>> Xt # doctest: +SKIP array([[ 0., 0., 0., 0.], [ 1., 1., 1., 0.], [ 2., 2., 2., 1.], [ 2., 2., 2., 2.]]) Sometimes it may be useful to convert the data back into the original feature space. The ``inverse_transform`` function converts the binned data into the original feature space. Each value will be equal to the mean of the two bin edges. >>> est.bin_edges_[0] array([-2., -1., 0., 1.]) >>> est.inverse_transform(Xt) array([[-1.5, 1.5, -3.5, -0.5], [-0.5, 2.5, -2.5, -0.5], [ 0.5, 3.5, -1.5, 0.5], [ 0.5, 3.5, -1.5, 1.5]]) rNleft)closedz array-likeonehotz onehot-denseordinaluniformquantilekmeansrwarn random_staten_binsencodestrategydtype subsampler_parameter_constraints)rrrr rcCs(||_||_||_||_||_||_dSNr)selfrrrrr rr%I/tmp/pip-unpacked-wheel-zrfo1fqw/sklearn/preprocessing/_discretization.py__init__s zKBinsDiscretizer.__init__cCsR||j|dd}|jtjtjfkr0|j}n|j}|j\}}|jdkr|jdk r|jdkrt|dkrt dt qt |j }||jkr|j||jdd }t||}n(|jdkrt|jtrtd |jd |jd }||}tj|td} t|D]} |dd| f} | | } } | | krZt d | d || <ttj tjg| | <q|jdkrt| | || d | | <n|jdkrtdd|| d }tt| || | <n|jdkrddlm}t| | || d }|d d|dddddfd}||| |d d}| | dddfj!dddf}|"|d d|ddd| | <tj#| | | | f| | <|jdkrtj$| | tjddk}| | || | <t%| | d || krt d| t%| | d || <q| |_&||_'d|j(krNt)dd|j'D|j(dk|d|_*|j* td t%|j'f|S)a Fit the estimator. Parameters ---------- X : array-like of shape (n_samples, n_features) Data to be discretized. y : None Ignored. This parameter exists only for compatibility with :class:`~sklearn.pipeline.Pipeline`. Returns ------- self : object Returns the instance itself. numericrrNrgjAzIn version 1.3 onwards, subsample=2e5 will be used by default. Set subsample explicitly to silence this warning in the mean time. Set subsample=None to disable subsampling explicitly.F)sizereplacez"Invalid parameter for `strategy`: z6. `subsample` must be used with `strategy="quantile"`.rz3Feature %d is constant and will be replaced with 0.rrdrr)KMeans?)Z n_clustersinitZn_init)rr)Zto_beging:0yE>zqBins whose width are too small (i.e., <= 1e-8) in feature %d are removed. Consider decreasing the number of bins.rcSsg|]}t|qSr%)npZarange.0ir%r%r& #sz(KBinsDiscretizer.fit..) categoriesZ sparse_outputr)+Z_validate_params_validate_datarr1float64float32shaperr warningsr FutureWarningrrchoicer isinstancer ValueError_validate_n_binszerosobjectrangeminmaxarrayinfZlinspaceZasarrayZ percentileZclusterr-fitZcluster_centers_sortZr_Zediff1dlen bin_edges_n_bins_rr_encoder)r$XyZ output_dtypeZ n_samples n_featuresrngZ subsample_idxr bin_edgesjjcolumnZcol_minZcol_maxZ quantilesr-Z uniform_edgesr0kmZcentersmaskr%r%r&rHs             ($   zKBinsDiscretizer.fitcCs|j}t|tr tj||tdSt|tddd}|jdksH|jd|krPt d|dk||kB}t |d}|jddkrd d d |D}t d t j||S) z0Returns n_bins_, the number of bins per feature.r)TF)rcopyZ ensure_2drrz8n_bins must be a scalar or array of shape (n_features,).rz, css|]}t|VqdSr#)strr2r%r%r& <sz4KBinsDiscretizer._validate_n_bins..zk{} received an invalid number of bins at indices {}. Number of bins must be at least 2, and must be an int.)rr>rr1fullintr ndimr:r?wherejoinformatr__name__)r$rPZ orig_binsrZbad_nbins_valueZviolating_indicesindicesr%r%r&r@-s" z!KBinsDiscretizer._validate_n_binscCst||jdkrtjtjfn|j}|j|d|dd}|j}t|jdD]8}tj ||dd|dd|fdd|dd|f<qJ|j d kr|Sd}d |j kr|j j}|j|j _z|j |}W5||j _X|S) a Discretize the data. Parameters ---------- X : array-like of shape (n_samples, n_features) Data to be discretized. Returns ------- Xt : {ndarray, sparse matrix}, dtype={np.float32, np.float64} Data in the binned space. Will be a sparse matrix if `self.encode='onehot'` and ndarray otherwise. NTF)rWrresetrr.right)Zsiderr) r rr1r8r9r7rKrCr:Z searchsortedrrM transform)r$rNrXtrRrSZ dtype_initZXt_encr%r%r&rdFs 6    zKBinsDiscretizer.transformcCst|d|jkr|j|}t|dtjtjfd}|jj d}|j d|krdt d ||j dt |D]P}|j |}|dd|ddd }|t|dd|f|dd|f<ql|S) a Transform discretized data back to original feature space. Note that this function does not regenerate the original data due to discretization rounding. Parameters ---------- Xt : array-like of shape (n_samples, n_features) Transformed data in the binned space. Returns ------- Xinv : ndarray, dtype={np.float32, np.float64} Data in the original feature space. rT)rWrrrz8Incorrect number of features. Expecting {}, received {}.Nr.r/)r rrMinverse_transformr r1r8r9rLr:r?r_rCrKint_)r$reZXinvrPrSrRZ bin_centersr%r%r&rfms"     (z"KBinsDiscretizer.inverse_transformcCs$t||}t|dr |j|S|S)aGet output feature names. Parameters ---------- input_features : array-like of str or None, default=None Input features. - If `input_features` is `None`, then `feature_names_in_` is used as feature names in. If `feature_names_in_` is not defined, then the following input feature names are generated: `["x0", "x1", ..., "x(n_features_in_ - 1)"]`. - If `input_features` is an array-like, then `input_features` must match `feature_names_in_` if `feature_names_in_` is defined. Returns ------- feature_names_out : ndarray of str objects Transformed feature names. rM)rhasattrrMget_feature_names_out)r$Zinput_featuresr%r%r&ris   z&KBinsDiscretizer.get_feature_names_out)r")N)N)r` __module__ __qualname____doc__r rr r typer1r8r9rr!dict__annotations__r'rHr@rdrfrir%r%r%r&rs2    t'&r)ZnumbersrZnumpyr1r;rbaserrZutils._param_validationrr r r Zutils.validationr r rrutilsrrr%r%r%r&s