U 3d\@sddlZddlZddlZddlZddlmZddlmZmZddl m Z ddl m Z ddl ZddlmZddlmZddlZdd lmZd d lmZdd lmZdd lmZddlmZddlmZddlmZddlmZddlmZddlm Z ddlm!Z!ddlm"Z"ddlm#Z#ddlm$Z$ddlm%Z%ddlm&Z&d dl'm(Z(d dl'm)Z)d dl'm*Z*d dl'm+Z+d dl'm,Z,d dl-m.Z.m/Z/m0Z0m1Z1m2Z2m3Z3d d l4m5Z5m6Z6m7Z7d d!l8m9Z9d d"l:m;Z;d d#l:mZ>d d%l?m@Z@d d&l?mAZAd d'l?mBZBd d(lCmDZDd d)lCmEZEd d*lFmGZGd d+lHmIZImJZJmKZKd d,lLmMZMd d-lLmNZNd d.lOmPZPd d/lQmRZRd d0lQmSZSd d1lQmTZTdd2lmUZUdd3lVmWZWmXZXdd4lYmZZZm[Z[d d5l\m]Z]d d6l\m^Z^d d7l_m`Z`maZambZbmcZcdadd8d9d:d;gZedd?Zge#ehd@dAdBZidCdDZjdEdFZkdGdHZldIdJZmdKdLZndMdNZodOdPZpdQdRZqdSdTZrdUdVZsdWdXZtd d[d\Zud]d^Zvd_d`ZwGdadbdbZxdcddZydedfZzdgdhZ{e#ehd@didjZ|e#ehd@dkdlZ}e#ehd@dmdnZ~e#ehd@dodpZe#ehd@d!drdsZdtduZe#ehefd@dvdwZdxdyZe#dzd{Zd|d}Ze#ehd@d~dZe#ehd@ddZddZe#ehd@ddZe#ehd@ddZe#ddZe#ddZe#ddZe#ehd@d"ddZe#ehd@ddZe#ehd@ddZddZe#ddZe#ddZe#ddZddZe#ehd@ddZe#ehd@ddZe#ddZe#ddZe#ehd@ddZe#ehd@ddZe#ehd@ddZe#ehd@d#ddZe#ehd@ddZe#ehd@ddZe#ehd@ddZe#d$ddZddZd%ddZddZe#ehd@ddZe#ehd@ddZe#ehd@ddÄZe#ehd@ddńZe#ehd@ddDŽZe#ehd@d&ddɄZe#dd˄Ze#ehd@dd̈́Ze#ddτZddфZddӄZe#ehd@ddՄZe#ehd@dYejfddׄZe#ddلZe#ehd@ddۄZe#ehd@dd݄Ze#ehd@dd߄Ze#ehd@ddZe#ehd@ddZe#ehd@ddZe#ehd@ddZe#ehd@ddZe#ehd@ddZddZddZeJfddZe#ehd@ddZe#ehd@ddZe#ehd@ddZe#ehd@ddZe#ehd@ddZe#ehd@ddZddZǐddZȐddZɐddZʐddZːdd Ze#ehd@d d Z͐d d ZΐddZϐddZАddZѐddZҐddZӐddZԐddZՐddZ֐ddZdS('N)deepcopy)partialwraps) signature)Real)sparse)rankdata)IS_PYPY)config_context)Interval _get_args)assert_raise_message)assert_array_equal)assert_array_almost_equal)assert_allclose)assert_allclose_dense_sparse)assert_array_less)set_random_state)SkipTest)ignore_warnings)create_memmap_backed_data)raises) is_scalar_nan)LinearRegression)LogisticRegressionRANSACRegressor)Ridge) SGDRegressor)clone ClusterMixin is_classifier is_regressoris_outlier_detectorRegressorMixin)accuracy_scoreadjusted_rand_scoref1_score)BaseRandomProjection) SelectKBest)SelectFromModel) make_pipeline)DataConversionWarning)NotFittedError)SkipTestWarning)train_test_split) ShuffleSplit) _safe_split) rbf_kernel linear_kernelpairwise_distances) sp_version) parse_version)check_is_fittedmake_constraint)generate_invalid_param_val)InvalidParameterError)shuffle) _DEFAULT_TAGS _safe_tags)has_fit_parameter _num_samples)StandardScaler)scale) load_iris make_blobsmake_multilabel_classificationmake_regressionZ PLSCanonicalZ PLSRegressionCCAZPLSSVDccs|jj}t|}tVtVtVt|drntVtVt V|dsnt Vt Vt t ddVt t ddVtVt tddV|dstVtVtV|tkrtV|d s|dstV|drtVtVt|d rtVtVtVtVdS) N sample_weightpairwiseones)kindzerosTreadonly_memmap no_validation allow_nansparsify) __class____name__rAcheck_no_attributes_set_in_initcheck_estimators_dtypescheck_fit_score_takes_yrB"check_sample_weights_pandas_series!check_sample_weights_not_an_arraycheck_sample_weights_listcheck_sample_weights_shape$check_sample_weights_not_overwrittenrcheck_sample_weights_invariance!check_estimators_fit_returns_selfcheck_complex_datacheck_dtype_object$check_estimators_empty_data_messagesCROSS_DECOMPOSITIONcheck_pipeline_consistencycheck_estimators_nan_infcheck_nonsquare_error!check_estimators_overwrite_paramshasattrcheck_sparsify_coefficientscheck_estimator_sparse_datacheck_estimators_pickle%check_estimator_get_tags_default_keys) estimatornametagsrqB/tmp/pip-unpacked-wheel-zrfo1fqw/sklearn/utils/estimator_checks.py _yield_checksUs@  rsccst|}tVtVtVtVtV|dr4tVtVttddVttdddVt V|dr~t Vt Vt Vt V|dstV|dstV|d rtVd |krtVtVtVdS) N multioutputTrPfloat32rQX_dtypeZ multilabelrRmultioutput_only requires_fit class_weight)rA"check_classifier_data_not_an_arraycheck_classifiers_one_label*check_classifiers_one_label_sample_weightscheck_classifiers_classes'check_estimators_partial_fit_n_featurescheck_classifier_multioutputcheck_classifiers_trainr#check_classifiers_regression_target6check_classifiers_multilabel_representation_invariance2check_classifiers_multilabel_output_format_predict8check_classifiers_multilabel_output_format_predict_probasz,_get_check_estimator_ids..z{}({})rT)Zprint_changed_onlyz\sN)callablerrrVkeywordsfuncjoinitemsrrir resubstr)objZkwstringrqrqrr_get_check_estimator_idsbs   rcCs t|dg}t|r|dgdgfkr|t|tr<|t}n>t|trR|t}n(t|trl|tdd}n|t dd}n|dgfkrt|tr|d td d fd tdd fgd }n"|d t d dfd t ddfgd }n&d|j d|}t |t t|n|}|S)z)Construct Estimator instance if possible._required_parametersrnZbase_estimatorr random_stater )C estimatorsZest1皙?alphaZest2)rzCan't instantiate estimator z parameters )getattrlen issubclassrrr'r r-r!rrVrrr1r) EstimatorZrequired_parametersrnmsgrqrqrr_construct_instances4             rcCs8t||\}}|s||fS|j|||jj|ddSdS)N)reason)Zmarks)_should_be_skipped_or_markedparammarkZxfail)rnrpytestZshould_be_markedr rqrqrr_maybe_mark_xfailsrcsJt|\}|s|St|tr(|jjn|jt|fdd}|S)Ncs"tddjjddS)Nz Skipping z for z: )rrUrV)argskwargs check_namernr rqrrwrappedsz_maybe_skip..wrapped)r rrrrVr)rnrZshould_be_skippedrrqrrr _maybe_skipsrcCs@t|tr|jjn|j}t|ddp&i}||kr>> from sklearn.utils.estimator_checks import parametrize_with_checks >>> from sklearn.linear_model import LogisticRegression >>> from sklearn.tree import DecisionTreeRegressor >>> @parametrize_with_checks([LogisticRegression(), ... DecisionTreeRegressor()]) ... def test_sklearn_compatible_estimator(estimator, check): ... check(estimator) rNcss|]}t|tVqdSN)rtype)restrqrqrr sz*parametrize_with_checks..uPassing a class was deprecated in version 0.23 and isn't supported anymore from 0.24.Please pass an instance instead.c3s>D]4}t|j}t|D]}t||}t||VqqdSr)rrVrrr)rnrorrr rqrrchecks_generators    z1parametrize_with_checks..checks_generatorzestimator, check)ids)r any TypeErrorr Z parametrizer)rrrrqrrrparametrize_with_checkss(rF deprecatedc sdkr|dkrd}t||dkr8d}t|t|ttrNd}t|tjfdd}|rp|S|D]H\}z |Wqvtk r}ztt |t W5d}~XYqvXqvdS)aCheck if estimator adheres to scikit-learn conventions. This function will run an extensive test-suite for input validation, shapes, etc, making sure that the estimator complies with `scikit-learn` conventions as detailed in :ref:`rolling_your_own_estimator`. Additional tests for classifiers, regressors, clustering or transformers will be run if the Estimator class inherits from the corresponding mixin from sklearn.base. Setting `generate_only=True` returns a generator that yields (estimator, check) tuples where the check can be called independently from each other, i.e. `check(estimator)`. This allows all checks to be run independently and report the checks that are failing. scikit-learn provides a pytest specific decorator, :func:`~sklearn.utils.parametrize_with_checks`, making it easier to test multiple estimators. Parameters ---------- estimator : estimator object Estimator instance to check. .. versionadded:: 1.1 Passing a class was deprecated in version 0.23, and support for classes was removed in 0.24. generate_only : bool, default=False When `False`, checks are evaluated when `check_estimator` is called. When `True`, `check_estimator` returns a generator that yields (estimator, check) tuples. The check is run by calling `check(estimator)`. .. versionadded:: 0.22 Estimator : estimator object Estimator instance to check. .. deprecated:: 1.1 ``Estimator`` was deprecated in favor of ``estimator`` in version 1.1 and will be removed in version 1.3. Returns ------- checks_generator : generator Generator that yields (estimator, check) tuples. Returned when `generate_only=True`. See Also -------- parametrize_with_checks : Pytest specific decorator for parametrizing estimator checks. Nr zBEither estimator or Estimator should be passed to check_estimator.ze'Estimator' was deprecated in favor of 'estimator' in version 1.1 and will be removed in version 1.3.rc3s,tD]}t|}t|fVqdSr)rrr)rrnrorqrrrps  z)check_estimator..checks_generator) rrr FutureWarningrrrrVrrr1)rnZ generate_onlyrrrr exceptionrqr!rrcheck_estimator$s*7    r$cCs8tdkr4tddddddd\}}t|}||fatS)Nrr @*) n_samples n_featuresZ n_informativeZbiasZnoiser)REGRESSION_DATASETrIrD fit_transformrrrqrqrr_regression_datasets  r.cCs|}|jj}|dkr$|jddd|kr@|dkr@|jddd|kr|jdk rf|jtd|jd|d krz|jd d|d kr|jd d|jjd kr|jd dd|dkr|jdd|dkr|jddd|kr|jddd|kr|jtd|jdd|kr|jddd|kr.|jddd|krP|dsP|jdd|dkrf|jd d!|d"krvd#|_|d$kr|jd d%t |d&rt|j d|_ t |d'rd#|_ |d(kr|jd)d*|d+krd|_ t |tr|jdd,t |tr|jd#d-|d.kr$|jdd/|d0kr:|jd1d2d3d4g}||krdt |d5rd|jd6d7t |d8r||jd6d9|d:kr|jd;d<|d=krttd>krd?nd@}|j|dA|tkr|jd#d,|dBkr|jdCdDdS)ENTSNEr Z perplexityn_iterrr1max_iterr3)Z LinearSVR LinearSVCr'ZNMFiZ MiniBatchNMFT)r3Zfresh_restarts)Z MLPClassifierZ MLPRegressordZMiniBatchDictionaryLearning n_resampling)r7 n_estimators)r8 max_trialsr)r9n_init)r: batch_sizeZMLP)r;Z MeanShift?) bandwidthZ TruncatedSVDr Z LassoLarsIC)Znoise_variance n_clustersn_bestZ SelectFdr?rZTheilSenRegressor) n_components)r)ZHistGradientBoostingClassifierZHistGradientBoostingRegressor)Zmin_samples_leafZDummyClassifierZ stratified)ZstrategyZRidgeCVZRidgeClassifierCVcvrBn_splits)rEZ OneHotEncoderignore)Zhandle_unknownZQuantileRegressorz1.6.0Zhighszinterior-point)solverZSpectralEmbeddinggh㈵>)Z eigen_tol)rrUrV set_paramsr3minr8rrArir>r?Zmax_subpopulationrr+r,r8r9rd)rnparamsroZloo_cvrGrqrqrr_set_checking_parameterss                                           rKc@s*eZdZdZddZd ddZddZdS) _NotAnArrayzvAn object that is convertible to an array. Parameters ---------- data : array-like The data. cCst||_dSr)rasarraydata)selfrNrqrqrr__init__sz_NotAnArray.__init__NcCs|jSr)rN)rOdtyperqrqrr __array__sz_NotAnArray.__array__cCs"|jdkrdStd|jdS)NZmay_share_memoryTz%Don't want to call array_function {}!)rVrr)rOrtypesrrrqrqrr__array_function__s z_NotAnArray.__array_function__)N)rVr __qualname____doc__rPrRrTrqrqrqrrrLs rLcCst|dd}t|dkS)zReturns True if estimator accepts pairwise metric. Parameters ---------- estimator : object Estimator object to test. Returns ------- out : bool True if _pairwise is set to True and False otherwise. metricNZ precomputed)rbool)rnrWrqrqrr_is_pairwise_metrics rYccs|jdkstd|fVdD]}|||fVq |d}|jd|_|jd|_d|fVdD]8}||}|jd|_|jd|_|d|fVqjdS) a;Generate sparse matrices with {32,64}bit indices of diverse format. Parameters ---------- X_csr: CSR Matrix Input matrix in CSR format. Returns ------- out: iter(Matrices) In format['dok', 'lil', 'dia', 'bsr', 'csr', 'csc', 'coo', 'coo_64', 'csc_64', 'csr_64'] csr)ZdokZlilZdiaZbsrcsccoor\int64Zcoo_64)r[rZZ_64N) rAssertionErrorcopyZasformatrowastypecolindicesZindptr)X_csrZ sparse_formatZX_coorrqrqrr_generate_sparse_matrix-s   rec Cstjd}|jdd}d||dk<t||}t|}d|jddt}t t dt |}W5QRXt ||}t |}t|D]J\}}t t d"t |}|dkr|jd d W5QRXd |krd |d |d} n d |d} tttfddgd| dt t d|||W5QRXt|drz||} |drb| j|jddfksztn| j|jdfksztt|dr||} |dr|jddf} n|jddf} | j| kstW5QRXqdS)Nr)(rCrg?rfr)ZScalerrDF)Z with_meanZ64rz doesn't seem to support z_ matrix, and is not failing gracefully, e.g. by using check_array(X, accept_large_sparse=False)z doesn't seem to fail gracefully on sparse data: error message should state explicitly that sparse input is not supported if this is not the case.rZSparseTrmay_passrrrxr predict_proba binary_onlyr )rrruniform_enforce_estimator_tags_XrZ csr_matrixraintrr"r"rrArerHrrrrrirshaper^rj) rorrrrdrrnrpZ matrix_formatrpredZprobsZexpected_probs_shaperqrqrrrkNsN                rkcCst|}zddl}tddgddgddgddgddgddgddgddgddgddgddgddgg }|t||}|ddddddddddddg }|dgd}t|ddr||}z|j|||d Wn"t k rt d |YnXWnt k rt d YnXdS) Nrr r rCrg rxrrKzPEstimator {0} raises error if 'sample_weight' parameter is of type pandas.SerieszUpandas is not installed: not testing for input of type pandas.Series to class weight.) r"pandasrarray DataFramermSeriesrArrr ImportErrorr)rorrnpdrrweightsrqrqrrrZsF"  rZcCst|}tddgddgddgddgddgddgddgddgddgddgddgddgg }tt||}tddddddddddddg }tdgd}t|ddrt|jdd}|j|||d dS) Nr r rCrgrqrxrrr) r"rrtrLrmrArNreshaper)rorrnrrryrqrqrrr[s,  r[cCsdt|}tjd}d}t||j|dfd}t|d}t||}dg|}|j|||ddS)NrrCrrr) r"rrrrmrlarangerr)rorrnrndr)rrrKrqrqrrr\s   r\cCs(t|}tddgddgddgddgddgddgddgddgddgddgddgddgddgddgddgddgg}tddddddddddddddddg}t||}|j||tt|dtt$|j||tdt|dW5QRXtt$|j||tt|dfdW5QRXdS)Nr rCr rgrr) r"rrtrrrMrrr)rorrnrrrqrqrrr]s6*  ( r]rMcCst|}t|}t|ddt|ddtjddgddgddgddgddgddgddgddgddgddgddgddgddgddgddgddggtjd}tjddddddddddddddddgtd}|dkr|}|}tjt|d } d |d } nz|d krjt||dg}t |d|g}tjt|dd } d| t|d<t ||| dd\}}} d |d } nt t ||}t ||}|j ||dd|j ||| ddD]<} t|| rt|| |} t|| |} t| | | dqdS)Nrrr rCr rgrQrM)rozFor z; sample_weight=None is not equivalent to sample_weight=onesrOz?, a zero sample_weight is not equivalent to removing the sample)rrK)rrjdecision_functionrr)r"rrrtfloat64rnrMrvstackZhstackr?rrrrirr)rorrNZ estimator1Z estimator2ZX1y1ZX2y2Zsw2rmethodX_pred1X_pred2rqrqrrr_sb  .      r_cCst|}t|ddtjddgddgddgddgddgddgddgddgddgddgddgddgddgddgddgddggtjd}tjddddddddddddddddgtd}t||}t|jd}d|d<| }|j |||d |d }t |||d dS) Nrrr rCr rgrg$@rrz8 overwrote the original `sample_weight` given during fitr) r"rrrtrrnrrMror_rr)rorrnrrZsample_weight_originalZsample_weight_fitrrqrqrrr^Fs< .  r^c Cstjd}t||jdd}|t}t|}|dddfdt}t |}t ||}| ||t |dr|| |t |dr||ttddd | ||tW5QRXd |d krd d i|d<d}tt|d| ||W5QRXn | ||dS)Nr)rfrrrgrrzUnknown label typeTrristringrZfoobarrrz"argument must be a string.* numberr)rrrrmrlraobjectrArnr"rrrirrr Exceptionr)rorrrrprrnrrqrqrrrbns(          rbc Cstjd}|jddd|jdd}|dd}|jdddd d}t|}t|dd tt d d | ||W5QRXdS) Nr(rry?rzr rr lowhighrrzComplex data not supportedr) rrrrlr{randintr"rrrr)rorrrrrnrqrqrrras   racCs|dkr dStjd}|dkr2d|jdd}nd|jdd}t||}|dddft}t|}t||}t |drd |_ t |d rd |_ t |d rd |_ t |d |||d D]<}t ||r|j}t||||j|kstd |qdS)N)ZSpectralCoclusteringrrrCr'rCrr rAr r>r?rrrrjz$Estimator changes __dict__ during %s)rrrrlrmrarnr"rrirAr>r?rr__dict__r_rr^)rorr~rrrnrZ dict_beforerqrqrrrs2           rcCs|dp|d S)N_)rr)attrrqrqrr_is_public_parametersrc st|jdrdSt|}tjd}d|jdd}t||}|dddft }t ||}t|drnd|_ t|dr~d|_ t |d|j||||jd d D}fd d |D}|rtd d |fdd |D}|r tdd |dS)Ndeprecated_originalrrCrrrAr r>cSsg|]}t|r|qSrq)rrrrqrqrrrsz3check_dont_overwrite_parameters..csg|]}|kr|qSrq)rr)dict_before_fitrqrrrs zEstimator adds public attribute(s) during the fit method. Estimators are only allowed to add private attributes either started with _ or ended with _ but %s added, cs g|]}||k r|qSrqrqrZdict_after_fitrrqrrrszEstimator changes public attribute(s) during the fit method. Estimators are only allowed to change attributes started or ended with _, but %s changed)rirPr"rrrrlrmrarnrrAr>rrr_rrr^r) rorrnr~rrZpublic_keys_after_fitZattrs_added_by_fitZattrs_changed_by_fitrqrrrrsF           rcCstjd}d|jdd}t||}|dddft}t|}t||}t |dr^d|_ t |drnd|_ t |d| ||dD]&}t ||rttd t|||dqdS) NrrCrrrAr r>rzReshape your data)rrrrlrmrarnr"rrirAr>rrrrr)rorr~rrrnrrqrqrrrs(        rcs~|}|jdfdd|D}t|tkrL|d}ttdd|}t|rj|j}dd|D}t |t |fS)Nr csg|]}|dqS)r )r{)rbatchrr*rqrrr"sz%_apply_on_subsets..rcSs|dS)Nrrq)xrqrqrr'z#_apply_on_subsets..cSsg|] }|jqSrq)A)rrrqrqrrr+s) rortuplelistmaprissparserrravel)rr result_fullresult_by_batchrqrrr_apply_on_subsetss   rc Cstjd}d|jdd}t||}|dddft}t|}t||}t |dr^d|_ t |drnd|_ t |d| ||dD]@}d j||d }t ||rtt|||\}} t|| d |d qdS) NrrCrrrAr r>rrr score_samplesrjz={method} of {name} is not invariant when applied to a subset.rroHz>atolr)rrrrlrmrarnr"rrirAr>rrrrrr) rorr~rrrnrrrrrqrqrrr0s.        rc Cstjd}d|jdd}t||}|dddftj}t|ddrVd||dk<t|}t ||}t |d rxd|_ t |d rd|_ t |d|||tj|jd}d D]H}d j||d }t ||rtt||||t||||d|dqdS)NrrCrrrkrr r rAr>rzY{method} of {name} is not invariant when applied to a datasetwith different sample order.r& .>r)rrrrlrmrar]rAr"rrirAr>rrZ permutationrorrr) rorr~rrrnidxrrrqrqrrrVs8          rc Cstjd}d|jdd}t||}|dddft}t|}t||}t |dr^d|_ t |drnd|_ t |d|dkr|j dd |d kr|j d d d dddddg}tt|dd|||W5QRXdS)NrrC)r rrrAr r>ZOPTICS)Z min_samplesr/r@r0z1 samplez n_samples = 1z n_samples=1z one samplez1 classz one classTr)rrrrlrmrarnr"rrirAr>rrHrrrrorr~rrrnZmsgsrqrqrrrs0         rc Cstjd}d|jdd}t||}|dddft}t|}t||}t |dr^d|_ t |drnd|_ |dkr|d|_ |d krd |_ t||}t|dd d d g}tt|dd|||W5QRXdS)NrrC)rr rrAr r>ZRandomizedLogisticRegressionrr@z1 feature\(s\)zn_features = 1z n_features=1Tr)rrrrlrmrarnr"rrirAr>Zsample_fractionZresidual_thresholdrrrrrrqrqrrrs&        rc Cstjd}d|jdd}|t}t|}t||}t|drHd|_ t|drXd|_ t |dt t |||W5QRXdS)NrrCr'rrAr r>)rrrrlrarnr"rrirAr>rrrr)rorr~rrrnrqrqrrrs       rcCsbtddddgdddggdddd\}}t|}t||}|rPt||g\}}t||||dSNr|rr r rr)centersrr* cluster_std)rGrDr,rmr_check_transformer)rorrQrrrqrqrrrs   rcCsztddddgdddggdddd\}}t|}t||}t|}tt|}t||||t||||dSr) rGrDr,rmrLrrMrtolist)rorrrZthis_XZthis_yrqrqrrrs   rc CsDt\}}t|}tttfd|dd||W5QRXdS)NzThe unfitted transformer z\ does not raise an error when transform is called. Perhaps use check_is_fitted in transform.r)r.r"rAttributeErrorrr)rorrrrqrqrrrs   rc CsTt|j\}}t|}t||tkrptjt|t|f}|ddddfd9<t|trtt|}n|}| ||t|}|j ||d} t| t r| D]} | jd|kst qn| jd|kst t |drP|tkr|||} |j ||d} n||} |j ||d} t|ddr8|d} t| t| t rt| t rt| | | D]4\} }}t| |d d |d t| |d d |d q\nLt| | d |d d t| | d d |d t| |kst t| |kst t |drPt|ddsP|jdkrP|jddkrPttd|dd ||ddddfW5QRXdS)Nr r rrrrr is non deterministic{Gz?z9fit_transform and transform outcomes not consistent in %srz7consecutive fit_transform outcomes not consistent in %s)rrrorzThe transformer zr does not raise an error when the number of features in transform is different from the number of features in fit.rrz)rrMror"rrdc_rrLrr,rr^rirrArziprrCndimrr)rotransformer_origrrr)r*ry_Ztransformer_cloneZX_predZx_predrZX_pred3rZx_pred2Zx_pred3rqrqrrrs            rc Cst|ddr|d}t|tddddgdddggdddd \}}t||td }t|}t||}t|t|}| ||| ||d d g}|D]@}t ||d} | dk rt ||} | ||} | ||} t | | qdS) Nrrrr|rr r rrkernelscorer,) rArrGrmr5r"rrr.rrr) rorrrrrnpipelinefuncs func_namerZ func_pipelineresultZ result_piperqrqrrrens2         rec Cstjd}d}|j|dfd}t||}t|d}t|}t||}t|ddddd g}|D]v}t ||d} | dk rd| ||d d t | j D} | dd kr| d d} | d dksdt d|t|j| fqddS)Nrr|rCrrr partial_fitrr,cSsg|] }|jqSrq)rorprqrqrrrsz+check_fit_score_takes_y..rOr )rYzJExpected y or Y as second argument for method %s of %s. Got arguments: %r.)rrrrlrmr}r"rrrr parametersvaluesr^rrV) rorr~r)rrrnrrrrrqrqrrrYs*       rYc Cstjd}d|jddtj}t||}|tj}|tj}|tj }|dddf}t ||}ddddg}||||fD]D} t |} t | d | | ||D]} t| | rt| | | qqdS) NrrCr'rrrrrrjr )rrrrlrarurmrr]int32rr"rrrir) rorr~Z X_train_32Z X_train_64ZX_train_int_64ZX_train_int_32rmethodsX_trainrnrrqrqrrrXs           rXc Cstddddgdddggddd\}}t|}t||}t|ddD]}||}t|}t||||}||| |}t ||gdd gD]H\} } t | t r| d} | j |kst|d | d |jd | j d qqFdS)Nr|rr r)r)rrrrrr,rz (method=z3) does not preserve dtype. Original/Expected dtype=z , got dtype=.)rGrDr,rmrArar"rrrrrrrQr^rV) rorrrrQZX_castrZX_trans1ZX_trans2ZXtrrqrqrrrs(       rcCst|}t|dtddd}d|d}tt|d||gW5QRXtddd}t|t ddddddddddddg }d}tt|d |||W5QRXdS) Nr rrCThe estimator zc does not raise a ValueError when an empty data is used to train. Perhaps use check_array in train.rrqzG0 feature\(s\) \(shape=\(\d*, 0\)\) while a minimum of \d* is required.r) r"rremptyr{rrrrrt)roreZX_zero_samplesrZX_zero_featuresrrrqrqrrrcs  (rcc Csftjd}t||jdd}|jdd}tj|d<|jdd}tj|d<td}d|dd<t||}d|d}d|d }d|d } ||fD]} t t d t |} t | d t td dg|d| | |W5QRX| ||t| dr"t td dg|d| | W5QRXt| drVt td dg| d| | W5QRXW5QRXqdS)NrrrCrrrrrz& doesn't check for NaN and inf in fit.z* doesn't check for NaN and inf in predict.z, doesn't check for NaN and inf in transform.rr rNaNrrr)rrrrmrlrrrMrrr"r"rrrrrirr) rorr~ZX_train_finiteZ X_train_nanZ X_train_infrerror_string_fiterror_string_predictZerror_string_transformrrnrqrqrrrfsJ                  rfc CsHtddd\}}t|}ttd|dd|||W5QRXdS)z8Test that error is thrown when non-square data provided.r'r)r)r*zThe pairwise estimator z+ does not raise an error on non-square datarN)rGr"rrrrorrrrnrqrqrrrg-s rgcCs>ddddg}tddddgdddggddd d \}}t||td }t|}|d r~tjd }|j|jddd}tj | d|<t |}t ||}t ||||t|} |j} | drd| ks| dsd| kstt| } t} |D] } t|| rt|| || | <q| D]"} t| | |}t| | |qdS)z'Test that we can pickle all estimators.rrrrjr|rr r rrrrSr(rF)replacerzrrrsversionN)rGrmr5rArrrchoicerrr{r"rrrpickledumpsrrrr^loadsdictrirr)ror check_methodsrrrprmaskrnZpickled_estimatorrZunpickled_estimatorrrZunpickled_resultrqrqrrrl=sD          rlc Cst|dsdSt|}tddd\}}t||}t||}z4t|r`t|}|j|||dn |||Wnt k rYdSXt t d|dd"||dddd f|W5QRXdS) Nr2r r)rclassesrzZ does not raise an error when the number of features changes between calls to partial_fit.rrz) rir"rGrmrr$runiquerNotImplementedErrorrr)rorrnrrrrqrqrrrps$     rcCs&d\}}}t|}t|}td|||d\}}|||||}|j||fkshtd||f|j|jj dksxtt |dr| |} t | t jst| j||fkstd||f| j| dkt} |j| } t| |t |d r||} t | trx|d sxt|D]^} | | j|d fksHtd |d f| | jtt j| | d dt|dd| fqnD|d s| j||fkstd ||f| jt| t|t |dr"t |d r"t|D]D} ||dd| f}| |}tt|t|dd| fqdS)N)r(rrCr()rr)n_labels n_classesSThe shape of the prediction for multioutput data is incorrect. Expected {}, got {}.irzaThe shape of the decision function output for multioutput data is incorrect. Expected {}, got {}.rrj poor_scorer zTThe shape of the probability for multioutput data is incorrect. Expected {}, got {}.r Zaxis)rAr"rHrrror^rrQrNrirrrndarrayrarnclasses_rrjrrangeargmaxroundr)rornr)rrrprry_preddecisiondec_preddec_expy_probrZy_probaZ y_decisionrqrqrrrsr               rcCst|}d}}t|s |d}tdd||d\}}t||}|||||}|jtdksttd |j|j |j kstddS) Nrr r(r)r n_targetsr)r*rzbMultioutput predictions by a regressor are expected to be floating-point precision. Got {} insteadr) r"rYrIrmrrrQrr^rro)rornr)r*rrrrqrqrrrs*    rc Cst|}tddd\}}t||dd\}}t|}tjd}t||j dddd g}|rtt |||g\}}}|j \}} t |d r|j dd t||d kr|j d d|j dd|||||j} | j |fkstt| |dkstt|ddrdSt|tjdd||} W5QRXt| | | jtdtdfksbt| jtdtdfkst||} t| } t| t| d| dd| ddkstt |d rt|d }|d| dkstdS)Nrr rrrC)rr rr>)r>ZAffinityPropagationi)Z preferencer6r4g?rrTrecordrr]rrz)rrz)r"rGr?rDr,rrrZ concatenaterlrrorirHrrrZlabels_r^r)rArcatch_warningsrrrQrr}r)roclusterer_origrQrrrrZX_noiser)r*rppred2labelsZ labels_sortedr>rqrqrrrsL               rcCsdtddd\}}t|}t|t|dr`|||}|jdd|||}t||dS)z2Check that predict is invariant of compute_labels.r'rrcompute_labelsF)rN)rGr"rrirrrHr)rorrrrrrrqrqrrr s  rc Csd}d}tjd}|jdd}|jdd}td}ttd^t|}tt dd |d } | ||W5QRX| j rW5QRdSt | |||d W5QRXdS) Nz6Classifier can't train when only one class is present.z8Classifier can't predict when only one class is present.rrrrrclassTrhr)rrrrlrMrr"r"rrrraised_and_matchedrr) roclassifier_origrrr~rX_testrrcmrqrqrrr|/s&     r|c Cs|d}|d}tjd}|jdd}|jdd}tdd}|}t|} t| drxd |g} tt f|} } nd } t t fd } } t | | d | d D} | j |||d| j rW5QRd St| |td|dW5QRXd S)zCheck that classifiers accepting sample_weight fit or throws a ValueError with an explicit message if the problem is reduced to one class. zu failed when fitted on one label after sample_weight trimming. Error message is not explicit, it should have 'class'.z; prediction results should only output the remaining class.r)rrrrr rKz\bclass(es)?\bz\bsample_weight\bNTrhrrr)rrrrlr}r_r"rBr^rrrrrrrrM)ror Z error_fitZ error_predictr~rr rrKrrZerr_typerr rqrqrrr}Fs.     r}rc Cs`tddd\}}||}t||dd\}}t|}||dk}||dk}|dkrp||8}||8}|rt||||g\}}}}||fg}t|} | ds|||f|D]\} } t | } t | } | j \}}t |}t|| } t|| } t|| d s@ttd |d d || | dd W5QRX|| | || | t|dspt|| }|j |fkst| dst| |dkstd}d}| d s&| drtt||dd || d dW5QRXn,tt||dd || jW5QRXt|drPz|| }| dkr| dsh|j |fks|tn|j |dfks|t|dkt}t||n(|j || fksttt j |dd|| d s6| dr tt||dd || d dW5QRXn,tt||dd || jW5QRXWnt!k rNYnXt|dr|"| }|j || fksxttt j |dd|t#t j$|ddt %|| d s| drtt||dd |"| d dW5QRXn,tt||dd |"| jW5QRXt|dr|&| }t'|t (|dddtt )|t )|qdS)N,rrrrr ) BernoulliNBZ MultinomialNB ComplementNBZ CategoricalNBrkrRThe classifier does not raise an error when incorrect/malformed input data for fit is passed. The number of training examples is not the same as the number of labels. Perhaps use check_X_y in fit.rrzrrg(\?zuThe classifier {} does not raise an error when shape of X in {} is not equal to (n_test_samples, n_training_samples)z|The classifier {} does not raise an error when the number of features in {} is different from the number of features in fit.rLrr rrxrrjpredict_log_probar)r)*rGrar?rDr,rIrrAappendrrrror"rmrrrrrrrir^rr(rr{TrrrnrrrrjrsumrMrrlogargsort)ror rQrwZX_mZy_mZy_bZX_bproblemsrprrrrr)r*rrZ msg_pairwiserrrrZ y_log_probrqrqrrrks                                     rcCsV||kr|}|d}n |}|d}t|}d}tt|||dksRt|dS)Nr zThe number of predicted outliers is not equal to the expected number of outliers and this difference is not explained by the number of ties in the decision_function values)rsortrrr^) num_outliersexpected_outliersrstartendZsorted_decisionrrqrqrrcheck_outlier_corruptions  rTc Csd}t|dd\}}t|dd}|r,t|}|j\}}t|}t|||||||}|j|fksxt |j j dkst t t |t ddg||} ||} | | fD](} | j t d kst | j|fkst qtt||jW5QRX| dkt} d| | dk<t | |tt||jW5QRX| |j} t| | tt||jW5QRXt|d rt|d sd }||}|j|d ||||}t |dk}||kr||} t||| dS)Nr rrrrrrzr floatrZnoveltyr|r)rGr?rror"rrrrr^rQrNrrrrtrrrrrrarnZoffset_rrirHrr)rorrQr)rrr*rnrrZscoresoutputrZy_decrrrrqrqrrr sP                   rcCst|dsdSd|jkrdS|jd}tdd|Ds@td|D]H}t|trD|jtkr|jdkr|j dkr|jdksD|j d ksDtd qDdS) N_parameter_constraintsrcSsg|]}t|tqSrq)rr )rcrqrqrrr` sz/check_outlier_contamination..zDcontamination constraints should contain a Real Interval constraint.r@r>rightneitherz:contamination constraint should be an interval in (0, 0.5]) rir"rr^rr rrleftr%closed)rorZcontamination_constraints constraintrqrqrrrT s.    rc Cstdddddddd\}}t|}|dd |dd }}|d d}|}t|}t|} t| | |||} | |||} | |||} t| | t| | | j | j kst | j | j kst t | t | kst t | t | kst dS) Nr6r rrCrTrr)r*rrlengthZallow_unlabeledrP) rHrErrr"rrrrrQr^r) ror rrry_trainr Zy_train_list_of_listsZy_train_list_of_arraysrrZy_pred_list_of_listsZy_pred_list_of_arraysrqrqrrro s:      rc Cs8t|}t|d\}}}t|d|ddddd\}}t|}|d| || d}} |d| || d} } ||| d } t|| d} | dkrt|d | d | | }t|tj st |d t |d |j | j kst |d|j d| j d|j | j ks4t |d|j d| j ddS)zeCheck the output of the `predict` method for classifiers supporting multilabel-indicator targets.r6rr rCrTrr*Nr does not have a  method.z2.predict is expected to output a NumPy array. Got instead.z(.predict outputs a NumPy array of shape instead of rz>.predict does not output the same dtype than the targets. Got )r"rrHrErrrrrrr^rrorQ)ror rr) test_size n_outputsrrrr r-y_testresponse_method_nameZpredict_methodrrqrqrrr s<    rc Cst|}t|d\}}}t|d|ddddd\}}t|}|d| || d}} |d| } ||| d } t|| d} | dkrt|d | d | | } t| trft | |kst d |d t | d|d| D]~}|j |dfkst d |d|j d|dfd|j j dks@t d |d|j dd |d}t|jddd|dqnt| tjr| j ||fkst d |d| j d||fd| j j dkst d |d| j dd |d}td| |dt| d|dntdt| d|ddS)zkCheck the output of the `predict_proba` method for classifiers supporting multilabel-indicator targets.r.r rCrTrr*Nrjr0r1zWhen zn.predict_proba returns a list, the list should be of length n_outputs and contain NumPy arrays. Got length of r3rzx.predict_proba returns a list, this list should contain NumPy arrays of shape (n_samples, 2). Got NumPy arrays of shape fzW.predict_proba returns a list, it should contain NumPy arrays with floating dtype. Got r2z.predict_proba returns a list, each NumPy array should contain probabilities for each class and thus each row should sum to 1 (or close to 1 due to numerical errors).r rrzX.predict_proba returns a NumPy array, the expected shape is (n_samples, n_outputs). Got zN.predict_proba returns a NumPy array, the expected data type is floating. Got z.predict_proba returns a NumPy array, this array is expected to provide probabilities of the positive class and should therefore contain values between 0 and 1.zUnknown returned type z by z4.predict_proba. A list or a Numpy array is expected.)r"rrHrErrrrrrr^rorQrNrrrrrrr)ror rr)r4r5rrrr r-r7Zpredict_proba_methodrrprrqrqrrr sb       rc Cs"t|}t|d\}}}t|d|ddddd\}}t|}|d| || d}} |d| } ||| d } t|| d} | dkrt|d | d | | } t| tj st |d t | d | j ||fkst |d| j d||fd| j jdkst |d| j d dS)zoCheck the output of the `decision_function` method for classifiers supporting multilabel-indicator targets.r.r rCrTrr*Nrr0r1z<.decision_function is expected to output a NumPy array. Got r2z].decision_function is expected to provide a NumPy array of shape (n_samples, n_outputs). Got r3rr8z?.decision_function is expected to output a floating dtype. Got )r"rrHrErrrrrrr^rrorQrN)ror rr)r4r5rrrr r-r7Zdecision_function_methodrrqrqrrr s<    rc Cs8t|}d|d}tt|d|W5QRXdS)zCheck the error raised by get_feature_names_out when called before fit. Unfitted estimators with get_feature_names_out should raise a NotFittedError. rzU should have raised a NotFitted error when fit is called before get_feature_names_outrN)r"rr0get_feature_names_out)rorrnrrqrqrr!check_get_feature_names_out_errorG s  r:cCs`tddd\}}t||}t|}t||}|r@t||g\}}t|||||ks\tdS)z+Check if self is returned when calling fit.rrr)N)rGrmr"rrrrr^)rorrQrrrnrqrqrrr`W s  r`c CsLt\}}t|}dD]0}t||rttt|||W5QRXqdS)z}Check that predict raises an exception in an unfitted estimator. Unfitted estimators should raise a NotFittedError. )rrrjrN)r.r"rirr0r)rorrrrnrrqrqrrrg s    rc Cs"t|}tjd}d}t||j|dfd}t|d}t||}t|}t || ||| |}t |t j dd8} t dtt dt| ||ddtjfW5QRX| |} d d d d | D} |d s t| dkst| d| ks tt|| dS)Nrr|rCrTralwaysrFz)expected 1 DataConversionWarning, got: %srcSsg|] }t|qSrqr)rZw_xrqrqrrr sz)check_supervised_y_2d..rtzPDataConversionWarning('A column-vector y was passed when a 1d array was expected)rArrrrmrlr}rr"rrrrr simplefilterr/RuntimeWarningZnewaxisrrr^rr) rorrpr~r)rrrnrwZ y_pred_2drrqrqrrr| s4      $   rc Cs|t|}t|}|dkr&||k}t||||||}t|dr$||}t |tj sjt t |dkr| dkt}|j|} t| |d|dtt| dtt|fdn`t|dd d kr$tj|d d t} |j| } t| |d|dtt| dtt|fd|d krDtt|t|t||jd |dtt|dtt|jfddS)Nr rr rzKdecision_function does not match classifier for %r: expected '%s', got '%s'rrZdecision_function_shapeZovrr rrz=Unexpected classes_ attribute for %r: expected '%s', got '%s')rrr"meanrrrrirrrr^rrrarnrrrrrrr) rrror rrrrrrZ decision_yZy_exprqrqrrcheck_classifiers_predictions s`           rCcCs|dkr |S|S)N)ZLabelPropagationZLabelSpreadingSelfTrainingClassifierrq)rory_namesrqrqrr _choose_check_classifiers_labels srFcCstdddd\}}t||dd\}}t|}||dk}||dk}t||}t||}dd d g}dd g}t||}t||} ||| fg} t|d d s| |||f| D]8\} } } | | d fD]}t || |}t | |||qqddg}t||} t ||| }t ||||dS)Nr|rr)r)rrrrr ZoneZtwoZthreerkrOrzr ) rGr?rDr,rmrZtakerArrarFrC)ror Z X_multiclassZ y_multiclassZX_binaryZy_binaryZlabels_multiclassZ labels_binaryZy_names_multiclassZy_names_binaryrrrrEZ y_names_irrqrqrrr~ s4              r~c Cst\}}t||dd}tjd}|jd|jdd}t||}tjd}t|}t|}t |t ||t krt |d||jdt |dg}|j }n|}|||||} |||t||} t| | d|ddS)NrrrCrr rr)r.rmrrrrrorr"rrdrrrrrrarr) roregressor_origrrr~rZ regressor_1Z regressor_2rpred1rrqrqrrr s&    $   rc Cs\t\}}||}t|}t|}t||}t||}|tkrxtj d}t |d||j dt |dg}|j }n|}|rt|||g\}}}t|dst|drd|_|dkrd|_ttd|d d |||dd W5QRXt|||||||||} | j|jks4tt|d d sX|||dksXtdS)Nrr ralphasrrZPassiveAggressiveRegressorrrrrzrrr@)r.rarEr"rmrrdrrrrrrrrrirrrrrrrrror^rAr) rorHrQrwrrrr~rrrqrqrrr) s:     $    rcCsrtjd}t|}|jdd}t||}t||dddf}|||dddg}|D]}t||rZt qZdS)Nr)rrgrrrjr) rrrr"normalrmrrrir^)rorHrrrrrrrqrqrrr\ s     rc CsNt|ddrdg}nddg}|D]&}t|ddd\}}t||ddd \}}}} t|d drpt||}t||}tt|}|dkrd d d } n d d d d} t|j| d} t | dr| jddt | dr| jd dt | dr| jddt | dr| jddt | | ||| |} t|dds t | dkdks tq dS)Nrkrr rCrr')rrrr@r4rrLg-C6?)rr )rr r rzr1r6r2r3r4min_weight_fraction_leafr)rOn_iter_no_change)rPrgףp= ?)rArGr2r5rrrr"rHrirrrrBr^) ror rZ n_centersrrrr r-r6rzrrrqrqrrrm s@                  rc Cst|}t|dr|jddt|dr4|jddt||||||}|jdd|||||} t|| d d t||d d kstdS) Nr1r6r2r3rMr4balancedrNZweighted)Zaverage)r"rirHrrrr*r^) ror rr-r r6ryrrZy_pred_balancedrqrqrr'check_class_weight_balanced_classifiers s          rRc Cs&tddgddgddgddgddgg}tdddddg}|}t|drX|jd d t|d rn|jd d t|d r|jddt||jdd|||j}t|}t tt |}|t |dk||t |dk|d}|j|d|||j} t || d|ddS)z4Test class weights with non-contiguous class labels.rgr<r$r rzr1rMr2r3r4rBrCrDrQrN)r rzz>Classifier %s is not computing class_weight=balanced properly.rN) rrtrirHrrcoef_r_rrrrr) ro ClassifierrrrZ coef_balancedr)rrzZ coef_manualrqrqrr-check_class_weight_balanced_linear_classifier s0(        rVc Cstddd\}}t||td}t|}t||}t||}t|}||||}| D]8\}} ||} t | t | ksdt d||| | fqddS)Nrr;r<rzTEstimator %s should not change or mutate the parameter %s from %s to %s during fit.) rGrmr5r"rrrrrrjoblibhashr^) rorrrrnrJZoriginal_params new_params param_nameoriginal_value new_valuerqrqrrrh s    rhcCsz t|}Wn$tk r0td|dYnXtt|jdrFdStt|j}trtdD]}||kr\||q\dddd t|jDD}t t |t |t |}|rt d |t |fdS) zCheck setting during init.rz9 should store all parameters as an attribute during init.rN)rcSsg|]}|D]}|q qSrqrq)rZ params_parentr rqrqrrr sz3check_no_attributes_set_in_init..css|]}t|VqdSrr)rparentrqrqrrr sz2check_no_attributes_set_in_init..zaEstimator %s should not set any attribute apart from parameters during init. Found attributes %s.) r"rrirrPrr remove__mro__setvarsr^sorted)rorrn init_paramsrZparents_init_paramsZ invalid_attrrqrqrrrW s,    rWc Cstddgddgddgddgddgddgddgddgddgg }tdddddddddg }t||}t|}|||||}|t|j st ||}t ||t t |}t|j st ||}t ||dS)Nrzr r rC)rrtrr"rrrTrrrTr^rrrr)rorrrrZ pred_origrprqrqrrrj s2       rjcCstddgddgddgddgddgddgddgddgddgddgddgddgg }t||}tddddddddddddg }t||}dD]}t|||||qdS)NrCrr r rg NotAnArrayZPandasDataframe)rrtrmr"check_estimators_data_not_an_arrayrorrrobj_typerqrqrrr{C s( " r{cCs<t\}}t||}t||}dD]}t|||||q"dS)Nre)r.rmrrgrhrqrqrrr\ s    rc Cs|tkrtdt|}t|}t|t||dkrFtd||dkrltt|}tt|}ndzFddl } t|}|j dkr| |}n | |}| t|}Wnt k rtdYnX|||||} |||||} t| | d|d dS) NzoSkipping check_estimators_data_not_an_array for cross decomposition module as estimators are not deterministic.rezData type {0} not supportedrfrr zDpandas is not installed: not checking estimators for pandas objects.rr)rdrr"rrrrLrrMrsrrvrurwrrr) rorrrriZ estimator_1Z estimator_2rZX_rxrIrrqrqrrrge s:         rgc s|j}ttdt|}t|t|||ks.param_filtercsg|]}|r|qSrqrqrrlrqrrr sz:check_parameters_default_constructible..rz(parameter %s for %s has no default valuez Parameter 'z' of estimator 'z ' is of type z which is not allowed. All init parameters have to be immutable to make cloning possible. Therefore we restrict the set of legal types to css|] }|jVqdSr)rV)rrrqrqrrr sz9check_parameters_default_constructible..rzEstimator parameter 'zT' is not returned by get_params. If it is deprecated, set its default value to None.z Parameter z> was mutated on init. All parameters must be stored unchanged.)*rUrr"rr"reprrHr^rrPrrrrrrrdefaultrrorrVrrnrrXrrS FunctionTyperWZMemoryupdatercoreZ numerictypesZallTypesr`rrrrr) rorrninitrcrJZ init_paramZ allowed_typesZ param_valueZ failure_textrqrmrrr sf     6     rcCstt|ddr |dt|7}t|ddrX|jdkrXt||jdk||jdd}t|ddrpt|dS|S)NZrequires_positive_yrr rkrrx)rzr )rAabsrIrrwhereZflatr{)rnrrqrqrrr s "  rcCsdt|ddkr |dddf}t|ddr8||}dt|ddkr\||tj}|jjdkrt||}t|rt|dd }nt|d dr|||}|S) NZ1darrayrrrr categoricalZSkewedChi2SamplerZ euclidean)rWrL) rArIrarrrUrVrYr7)rnrrrqrqrrrm s      rmcCsdddddddg}|t7}||kr&dS|dkr@t|jd d }nt|}t|d rt}|j|j}}t||}t|d t ||}| ||t |j d kstdS)Nr ZRidgeClassifierrrr5rrDZ LassoLarsr$rr3rr )rdr"rHrirFrNtargetrrrmrralln_iter_r^)rorZnot_run_check_n_iterrnZirisrrrqrqrrr s,       rcCst|}t|dr|tkr\dddgdddgdddgdddgg}ddgd d gdd gd dgg}n0td dddgdddggdddd\}}t||}t|d||||tkr|jD]}|dkstqn|jdkstdS)Nr3r$r<g@r&g@rgɿg?g?gg333333?r|rr r r) r"rirdrGrmrrryr^)rorrnrrZiter_rqrqrrrK s& $     rcsBt|}|jdd}|jddtfdd|Ds>tdS)NFdeepTc3s|]}|kVqdSr)r)ritemZ deep_paramsrqrrrr sz.check_get_params_invariance..)r"rrxrr^)rorrZshallow_paramsrqr}rrrj s  rc Cst|}|jdd}d}|jf||jdd}t|t|ksPt||D]\}}|||ksXt|qXtj tjdg}t |} |D]Z} || } |D]>} | | | <z|jf| Wnt t fk r} z| j j }td|| |d|}|}|jdd}zHt|t|ks4t|D]\}}|||ksget_params result does not match what was passed to set_paramszn{0} occurred during set_params of param {1} on {2}. It is recommended to delay parameter validation until fit.z9Estimator's parameters changed after set_params raised {})r"rrHr`rr^rrrrrrrUrVrrr)rorrnZ orig_paramsrZ curr_paramsrrZ test_valuesZ test_paramsrZ default_valuerre_typeZchange_warning_msgZparams_before_exceptionrqrqrrru sX        " "rc CsTt\}}t||}t|}d}t|ddsPtt|d|||W5QRXdS)NzUnknown label type: rRrr)r.rmr"rArrr)rorrrrrrqrqrrr s   rc sddg}tddd|ddd\}}t||d dd \}}}}t|} t| d r t| d r | ||| |ddd fjdd} | |jddt| t} zt | WnTt k rt fddt D} t | } t| t t| YnXdS)N)r r )rgrgr6rrgr<T)r)rr*rrr?皙?rLrrjr r)Zdecimalscsg|]}|kqSrq)rB)rgroupbZ rank_probarqrrr sz4check_decision_proba_consistency..)rGr2r"rirrjrrrrr^rrtrrrr}r)rorrrrrr r-r6rnaZ rank_scoreZgrouped_y_scoreZ sorted_idxrqrrrr s:   rc Csd}t|dd\}}t|dd}|j\}}t|}t|||}|j|fksTt|jjdksdtt t |t ddgt |d r|||}t ||t |d rd } t| |} |j| d ||}t |dk} | | krt |d r||} t| | | dS)Nr rrrrrrzr rrr|r r)rGr?ror"rrr^rQrNrrrrtrirrrrHrrr) rorr)rrr*rnrZy_pred_2rrrrrqrqrrr s2           rc CsPtddgddgg}tddg}t|}tt|||W5QRXdS)NrSr r )rrtr"rrrrrqrqrrrs  rc sddddg}tjd}t|tdkrDjddd }|jd |d fd }t |}t |rz|j|d }n|j dd |d }t |}t td|d|\}}t|||\} } t||||\} | | fdd|D} t| | |D]} t| r t| }t|jtjrPd t|jj}nd ttjj}t| | |t|dt|dd| dq dS)Nrrrrjr warm_startFrr6r locrrrrrLcs&i|]}t|r|t|qSrq)rir)rrr rnrqrr ?s z(check_fit_idempotent..rrz&Idempotency check failed for method {})rZrtolr)rrrr"rrrrHrKrmr%rrnextr3splitr4rrirZ issubdtyperQZfloatingZfinfoZepsrrmaxr)rorrrr)rrZtraintestrr-r6rrZ new_resultZtolrqrrrrsF         rc Cstjd}t|}t|d|kr4|jddd}|jd|dfd}t||}t |rj|j|d}n|j d d|d }t ||}t | d dszt|t|jjd Wntk rYnX|||z t|Wn.tk r}ztd |W5d}~XYnXdS)Nr(rFrr6r rrrrrz) passes check_is_fitted before being fit!zFEstimator fails to pass `check_is_fitted` even though it has been fit.)rrrr"rrrHrKrmr%rrrAgetr:r^rUrVr0r)rorrrnr)rrrrqrqrrrYs:        rcCstjd}t|}t|d|kr4|jddd}|jd|dfd}t||}t |rj|j|d}n|j dd|d }t ||}t |d rt |||t |d st |j|jd kst dS) NrrFrr6r rrrn_features_in_r )rrrr"rrrHrKrmr%rrrir^rrro)rorrrnr)rrrqrqrrrs       rc stjd}t|}t|d}|jd|dfd}t||}d}z||dWn<tk rzt fdd|Ds|W5dXYnXdS)Nrr6r r)z1requires y to be passed, but the target y is Nonez)rrverqrrrsz(check_requires_y_none..) rrrr"rrKrmrrr)rorrrnr)rZexpected_err_msgsrqrrrrs  rc Cst|}d|dkpd|dk}|r,|dr0dStjd}t|}t|d|krd|jddd }|j|d fd }t ||}t |r|j|d }n|j dd |d }t ||}| |||j|jdkstdddddg} |dddgf} d|jdd} | D]V} t|| sqt|| } | dkr.columnsr rerrorz#X does not have valid feature namesZsklearn)messagermodulefeature_names_in_zTEstimator does not have a feature_names_in_ attribute after fitting with a dataframerrrrz2 does not document its feature_names_in_ attribute)rrrrjrrrrrrzz.css|]}|dkVqdS)TNrq)rrrqrqrrrvsz;check_dataframe_column_names_consistency..zBThe feature names should match those that were passed during fit. z did not raiserrrr)2rsrwrrArrrr"rrKrmrortrrur%rrrrfilterwarnings UserWarningrrirrrrr^rQrrrrrrVrrrrIrrrrrescaperr$rr)rorrxrprrrnZX_origr)r*namesrrrrrrr invalid_namesrJZearly_stopping_enabledZ invalid_nameZadditional_messagerZ expected_msgrrqrqrr(check_dataframe_column_names_consistencys                          rc Cs|}d|dks|dr dStddddgdddggdddd \}}t|}t|}t||}|jd}t||}|tkrt j t |t |f}|ddddfd9<|j||d }d d t |D} t td d|| dddW5QRX|| } | dk stt| t js,t| jtks.z'input_features should have length equalrcss|]}t|tVqdSr)rr)rrorqrqrrrsz:check_transformer_get_feature_names_out.. Expected  feature names, got )rrGrDr,r"rmrorrdrrrMrrrr9r^rrrQrrxrr) rorrprrrr*r X_transformZinput_featuresZfeature_names_outn_features_outrqrqrr'check_transformer_get_feature_names_outsF        rc Csz ddl}Wntk r(tdYnX|}d|dksF|drJdStddddgdddggddd d \}}t|}t|}t||}|j d}t ||}|t krt j t |t |f}|ddddfd9<d d t|D} |j|| d } |j| |d} dd t|D} ttdd|| W5QRX|} || }t| |t| trz| dj d}n | j d}t| |kstd|dt| dS)NrrrrrRr|r r rrcSsg|]}d|qSrbrqrrqrqrrrszBcheck_transformer_get_feature_names_out_pandas..rrcSsg|]}d|qS)badrqrrqrqrrrsz0input_features is not equal to feature_names_in_rrr)rsrwrrrGrDr,r"rmrorrdrrrMrrurrr9rrrrr^)rorrxrprrrr*rfeature_names_indfrZinvalid_feature_namesZfeature_names_out_defaultZfeature_names_in_explicit_namesrrqrqrr.check_transformer_get_feature_names_out_pandassV          rc CsHtjd}|jdd}|jdddd}t||}|jdd}|r|j}t |t |}t |t |}d|d |d |} ||kst | t d d i} d dddg} |D]} |j| } | dkrqd| d|d}|d| d} t |}|j f| | i| D]l}t||s q tt|| dBtddt|ddDr\t|||nt||||W5QRXq dd| D} | D]}zt|| }Wntk rYqYnX|j f| |i| D]l}t||sqtt|| dBtddt|ddDr t|||nt||||W5QRXqАqqdS)Nrrrr r'Frzz>Mismatch between _parameter_constraints and the parameters of z%. Consider the unexpected parameters z% and expected but missing parameters ZBadTyperqrrr,rrRzThe 'z' parameter of z must be .* Got .* instead.z@ does not raise an informative error message when the parameter z% does not have a valid type or value.rcss"|]}t|to|dVqdSrN)rrrrZX_typerqrqrrr1sz)check_param_validation..rrcSsg|] }t|qSrqr;)rr)rqrqrrr=sz*check_param_validation..css|]}|dVqdSr)rrrqrqrrrMs)rrrrlrrrrr"r`r^rr"rHrirr>rrArr=r)rorrrrZestimator_paramsZvalidation_paramsZunexpected_paramsZmissing_paramsrZparam_with_bad_typeZ fit_methodsrZ constraintsrrnrr)Z bad_valuerqrqrrcheck_param_validations`            rc s|}d|dks|dr dStjd}t|}|jddt||jddddt|t |fd d }fd d }||g}|D]R}t|}||} t kr| d} |j d d||} t kr| d} t | | qdS)NrrrRrrrr r'cs.tkr|S|Sr)rdrrrrrorrqrrfit_then_transformgsz6check_set_output_transform..fit_then_transformcs |Sr)r,rr-rqrrr,lsz1check_set_output_transform..fit_transformror) rrrrr"rlrmrrrrd set_outputr) rorrprrrr,Ztransform_methodsZtransform_methodZX_trans_no_settingZX_trans_defaultrqrrrcheck_set_output_transformWs.     rc Csi}d||fd||fd||fd||fg}|D]J\}}} ||||tkr\|| |\} } n || } | |f||<q,d|fd|fg}|D]>\}} |tkr|| |\} } n || |} | |f||<q|S)zGenerate output to test `set_output` for different configuration: - calling either `fit.transform` or `fit_transform`; - passing either a dataframe or a numpy array to fit; - passing either a dataframe or a numpy array to transform. zfit.transform/df/dfzfit.transform/df/arrayzfit.transform/array/dfzfit.transform/array/arrayzfit_transform/dfzfit_transform/array)rrdrr9r,) rrorrroutputsZcasescaseZdata_fitZdata_transformX_transrrNrqrqrr_output_from_fit_transforms2     rc Csddl}|\}}|\}}t||js(t|j||d} z|j|| Wn<tk r} zt|d|d| | W5d} ~ XYnXdS)Nrrz, does not generate a valid dataframe in the z] case. The generated dataframe is not equal to the expected dataframe. The error message is: )rsrrur^ZtestingZassert_frame_equal) roroutputs_defaultoutputs_pandasrxrZfeature_names_defaultZdf_transZfeature_names_pandasZexpected_dataframerrqrqrr_check_generated_dataframesrc Cshz ddl}Wntk r(tdYnX|}d|dksF|drJdStjd}t|}|jdd}t ||}|j ddd d}t ||}t |d d t |jd D}|j||d } t|jdd} t| ||| |} t|jdd} zt| ||| |} Wn@tk r@}z t|dks*t|WYdSd}~XYnX| D]}t||| || |qFdS)NrrrrrRrrr r'cSsg|]}d|qSrrqrrqrqrrrsz5check_set_output_transform_pandas..r rrorrs+Pandas output does not support sparse data.)rsrwrrrrrr"rlrmrrrrrorurrrrr^rrorrxrprrrrrrZtransformer_defaultrZtransformer_pandasrrrrqrqrr!check_set_output_transform_pandassB      rc Csvz ddl}Wntk r(tdYnX|}d|dksF|drJdStjd}t|}|jdd}t ||}|j dd d d}t ||}t |d d t |jd D}|j||d} t|jdd} t| ||| |} t|} z*tddt| ||| |} W5QRXWn@tk rN}z t|dks8t|WYdSd}~XYnX| D]}t||| || |qTdS)z`Check that setting globally the output of a transformer to pandas lead to the right results.rNrrrrRrrr r'cSsg|]}d|qSrrqrrqrqrrrsz8check_global_ouptut_transform_pandas..r rrorrs)Ztransform_outputr)rsrwrrrrrr"rlrmrrrrrorurrr rrr^rrrqrqrr$check_global_ouptut_transform_pandassP       r)NFr )rM)F)F)Fr)T)F)rSrrrr_r functoolsrrinspectrZnumbersrZnumpyrZscipyrZ scipy.statsrrWrr r Z_param_validationr rrrrrrrrrrrrrrZ linear_modelrrrr r!baser"r#r$r%r&r'Zmetricsr(r)r*Zrandom_projectionr+Zfeature_selectionr,r-rr. exceptionsr/r0r1Zmodel_selectionr2r3Zmodel_selection._validationr4Zmetrics.pairwiser5r6r7Z utils.fixesr8r9Zutils.validationr:Zutils._param_validationr<r=r>r?Z_tagsr@rAZ validationrBrCZ preprocessingrDrEZdatasetsrFrGrHrIr+rdrsrr"rrrrrrrrrrr rr$r.rKrLrYrerkrZr[r\r]r_r^rrbrarrrrrrrrrrrrrrrerYrXrrcrfrgrlrrrrrr|r}rrrrrrrrr:r`rrrCrFr~rrrrrrRrVrhrWrjr{rrgrrrmrrrrrrrrrrrrrrmrrrrrrrrrrqrqrqrrs                                               4$ " 9&, > ]u!6 +   $D(  ! ' :  % + (    \      0  2  < ;   $   H ( , S -   # @ "  2  -  ( ! # "   .b  *   8  -+ : '  D   0 9 Y + +  )