-ie6SrSSKrSSKJrJr SSKJrJr SSK J r SSK J r J r SSKJr S S KJr S S KJr S S KJrJrJrJr S r"SS\\5rg)z This module contains the BinMapper class. BinMapper is used for mapping a real-valued dataset into integer-valued bins. Bin thresholds are computed with the quantiles so that each bin contains approximately the same number of samples. N) BaseEstimatorTransformerMixin) check_arraycheck_random_state)_openmp_effective_n_threads)Paralleldelayed)check_is_fitted) _map_to_bins)set_bitset_memoryview) ALMOST_INFX_BINNED_DTYPEX_BITSET_INNER_DTYPEX_DTYPEc[R"U5nUR5(aX)n[R"U5n[R"U5R [ 5n[U5U::aUSSUSS-nUS-nO^[R"SSUS-S9nUSSn[R"XSS 9R [ 5nURSUS- :Xde[R"US[US 9 U$) aExtract quantiles from a continuous feature. Missing values are ignored for finding the thresholds. Parameters ---------- col_data : array-like, shape (n_samples,) The continuous feature to bin. max_bins: int The maximum number of bins to use for non-missing values. If for a given feature the number of unique values is less than ``max_bins``, then those unique values will be used to compute the bin thresholds, instead of the quantiles Return ------ binning_thresholds : ndarray of shape(min(max_bins, n_unique_values) - 1,) The increasing numeric values that can be used to separate the bins. A given value x will be mapped into bin value i iff bining_thresholds[i - 1] < x <= binning_thresholds[i] Nr g?rd)nummidpoint)method)a_mina_maxout) npisnananysortuniqueastyperlenlinspace percentileshapeclipr)col_datamax_bins missing_maskdistinct_values midpoints percentiless c/var/www/html/venv/lib/python3.13/site-packages/sklearn/ensemble/_hist_gradient_boosting/binning.py_find_binning_thresholdsr.s.88H%LM*wwx Hii)009O ?x'#CR(?12+>> S kk!Shl; !!B' MM( KRR  q!X\111GGITC cP\rSrSrSrS\"S5SSSS4SjrS SjrSrS r S r g) _BinMapperKaTransformer that maps a dataset into integer-valued bins. For continuous features, the bins are created in a feature-wise fashion, using quantiles so that each bins contains approximately the same number of samples. For large datasets, quantiles are computed on a subset of the data to speed-up the binning, but the quantiles should remain stable. For categorical features, the raw categorical values are expected to be in [0, 254] (this is not validated here though) and each category corresponds to a bin. All categorical values must be known at initialization: transform() doesn't know how to bin unknown categorical values. Note that transform() is only used on non-training data in the case of early stopping. Features with a small number of values may be binned into less than ``n_bins`` bins. The last bin (at index ``n_bins - 1``) is always reserved for missing values. Parameters ---------- n_bins : int, default=256 The maximum number of bins to use (including the bin for missing values). Should be in [3, 256]. Non-missing values are binned on ``max_bins = n_bins - 1`` bins. The last bin is always reserved for missing values. If for a given feature the number of unique values is less than ``max_bins``, then those unique values will be used to compute the bin thresholds, instead of the quantiles. For categorical features indicated by ``is_categorical``, the docstring for ``is_categorical`` details on this procedure. subsample : int or None, default=2e5 If ``n_samples > subsample``, then ``sub_samples`` samples will be randomly chosen to compute the quantiles. If ``None``, the whole data is used. is_categorical : ndarray of bool of shape (n_features,), default=None Indicates categorical features. By default, all features are considered continuous. known_categories : list of {ndarray, None} of shape (n_features,), default=none For each categorical feature, the array indicates the set of unique categorical values. These should be the possible values over all the data, not just the training data. For continuous features, the corresponding entry should be None. random_state: int, RandomState instance or None, default=None Pseudo-random number generator to control the random sub-sampling. Pass an int for reproducible output across multiple function calls. See :term:`Glossary `. n_threads : int, default=None Number of OpenMP threads to use. `_openmp_effective_n_threads` is called to determine the effective number of threads use, which takes cgroups CPU quotes into account. See the docstring of `_openmp_effective_n_threads` for details. Attributes ---------- bin_thresholds_ : list of ndarray For each feature, each array indicates how to map a feature into a binned feature. The semantic and size depends on the nature of the feature: - for real-valued features, the array corresponds to the real-valued bin thresholds (the upper bound of each bin). There are ``max_bins - 1`` thresholds, where ``max_bins = n_bins - 1`` is the number of bins used for non-missing values. - for categorical features, the array is a map from a binned category value to the raw category value. The size of the array is equal to ``min(max_bins, category_cardinality)`` where we ignore missing values in the cardinality. n_bins_non_missing_ : ndarray, dtype=np.uint32 For each feature, gives the number of bins actually used for non-missing values. For features with a lot of unique values, this is equal to ``n_bins - 1``. is_categorical_ : ndarray of shape (n_features,), dtype=np.uint8 Indicator for categorical features. missing_values_bin_idx_ : np.uint8 The index of the bin where missing values are mapped. This is a constant across all features. This corresponds to the last bin, and it is always equal to ``n_bins - 1``. Note that if ``n_bins_non_missing_`` is less than ``n_bins - 1`` for a given feature, then there are empty (and unused) bins. gjANcLXlX lX0lX@lXPlX`lgN)n_bins subsampleis_categoricalknown_categories random_state n_threads)selfr6r7r8r9r:r;s r-__init___BinMapper.__init__s% ", 0("r/c^^^STRs=::aS::d&O [SRTR55e[T[/SS9mTRS- m[ TR 5nTRbUTRSTR:a8URTRSTRSS 9nTRUSS 9mTRc6[R"TRS[RS 9TlO2[R "TR[RS 9TlTRSnTR"nUcS/U-n[%U5HJnTRUnXgn U(aU c[S US 35eU(aM8U cM=[SUS35e TRS- TlS/U-TlS/U-n [+TR,SS9"UUU4Sj[%U555n Sn [%U5HtnTRU(a&Xgn U RSX'U TR(U'M=XTR(U'TR(URSS-X'U S- n Mv [R."U [R0S 9TlT$)a%Fit data X by computing the binning thresholds. The last bin is reserved for missing values, whether missing values are present in the data or not. Parameters ---------- X : array-like of shape (n_samples, n_features) The data to bin. y: None Ignored. Returns ------- self : object rr3z=n_bins={} should be no smaller than 3 and no larger than 256.Fdtypeensure_all_finiter Nr)replace)axisrAzKnown categories for feature z must be provided.zFeature zC isn't marked as a categorical feature, but categories were passed. threading)n_jobsbackendc3># UH8nTRU(aM[[5"TSS2U4T5v M: g7fr5)is_categorical_r r.).0f_idxXr(r<s r- !_BinMapper.fit..sBR *''. EG, -a5k8 D D*s A$A)r6 ValueErrorformatrrrr:r7r%choicetaker8rzerosuint8rJasarrayr9rangemissing_values_bin_idx_bin_thresholds_r r;arrayuint32n_bins_non_missing_)r<rMyrngsubset n_featuresr9rLr8 known_catsn_bins_non_missingnon_cat_thresholds non_cat_idx thresholdsr(s`` @r-fit_BinMapper.fits"T[['C'OVVKK  'e D;;? !2!23 >> %!''!*t~~*EZZ DNNEZJFvA&A    &#%88AGGAJbhh#GD #%::d.A.A#RD WWQZ 00  # $v 2 :&E!11%8N)0J*"4 3E7:LM">j&< ug&22'(,{{Q$ $v 2"Vj0%T^^[QR z*R    :&E##E* .4 ,6,<,bii#P  r/c[U[/SS9n[U5 URSURRS:wa?[ SR URRSURS55e[UR5n[R"U[SS9n[UURURURUU5 U$)aBin data X. Missing values will be mapped to the last bin. For categorical features, the mapping will be incorrect for unknown categories. Since the BinMapper is given known_categories of the entire training data (i.e. before the call to train_test_split() in case of early-stopping), this never happens. Parameters ---------- X : array-like of shape (n_samples, n_features) The data to bin. Returns ------- X_binned : array-like of shape (n_samples, n_features) The binned data (fortran-aligned). Fr@r rzKThis estimator was fitted with {} features but {} got passed to transform()F)rAorder)rrr r%r\rPrQrr;r zeros_likerr rYrJrX)r<rMr;binneds r- transform_BinMapper.transforms( 'e D 771:1177: :!!'(@(@(F(Fq(I177ST:!V  0? qcB    ( (     r/c[R"UR5nURRnURn[R"U[R S9n[R "U[R S9XA'URn[R"US4[S9n[U5HupxXXHn [XgU 5 M M Xd4$)aPCreate bitsets of known categories. Returns ------- - known_cat_bitsets : ndarray of shape (n_categorical_features, 8) Array of bitsets of known categories, for each categorical feature. - f_idx_map : ndarray of shape (n_features,) Map from original feature index to the corresponding index in the known_cat_bitsets array. rE) r flatnonzerorJsizerTr[arangerYr enumerater) r<categorical_features_indicesr`n_categorical_features f_idx_mapr9known_cat_bitsets mapped_f_idxrL raw_cat_vals r-make_known_categories_bitsets(_BinMapper.make_known_categories_bitsets+s(*~~d6J6J'K$)).. !=!B!BHHZryy9 24)) ""))3 / //HH #Q '/C  $--I#J L/6 %&7&E{S 7$K!++r/) rYr8rJr9rXr6r\r;r:r7r5) __name__ __module__ __qualname____firstlineno____doc__intr=rfrmr{__static_attributes__r/r-r1r1Ks8Ofc(# Tl&P",r/r1)rnumpyrbaserrutilsrrutils._openmp_helpersrutils.parallelr r utils.validationr _binningr _bitsetrcommonrrrrr.r1rr/r-rsA34@//"*MM0fB,!=B,r/