ri< dZddlZddlZddlmZmZddlZddlZddl Z ddl m Z ddl m Z ddlmZmZmZddlmZmZddlmZmZmZmZmZmZmZmZdd lmZdd l m!Z!dd l"m#Z#m$Z$dd l%m&Z&m'Z'm(Z(dd l)m*Z*m+Z+ddl,m-Z-ddl.m/Z/m0Z0ddl1m2Z2ddl3m4Z4m5Z5ddl6m7Z7m8Z8m9Z9m:Z:m;Z;ddlZ>ddl?m@Z@ddlAmBZBmCZCmDZDddlEmFZFmGZGe@dZHeZIeHjeIjjZMeIjeMeI_NeIjeMeI_KeZOeHjeOjjZMeOjeMeO_NeOjeMeO_KdZPe jjdeeGeFzdddddddddddddd dddd!ggd"d#ZSd$ZTe jjd%eGeFzd&ZUGd'd(e ZVd)ZWd*ZXd+ZYd,ZZd-Z[d.Z\d/Z]d0Z^e jjd1Z`d2Zae jjd3Zbd4Zcdbd5Zdd6Zed7Zfd8Zgd9Zhd:Zid;ZjGd<d=e ZkGd>d?e Zle jjd@eege jjdAddge jjdBddge jjdCdDdEgdFZmdGZndHZodIZpdJZqdKZrdLZsdMZtdNZudOZvdPZwdQZxdRZye jjdSeedTUdfeedTUdfee#dfee5dfgdVZze dWe jjdXeedTYdTZeedTYdTZgd[Z{e jjd\e9d]d]fe8d^d_fe7d^d]fge dWd`Z|e jjdXeedTYdTZeedTYdTZgdaZ}y)czE Testing for the bagging ensemble module (sklearn.ensemble.bagging). N)cycleproduct)config_context) BaseEstimator) load_diabetes load_irismake_hastie_10_2)DummyClassifierDummyRegressor)AdaBoostClassifierAdaBoostRegressorBaggingClassifierBaggingRegressorHistGradientBoostingClassifierHistGradientBoostingRegressorRandomForestClassifierRandomForestRegressor)_get_n_samples_bootstrap) SelectKBest)LogisticRegression Perceptron) GridSearchCV ParameterGridtrain_test_split)KNeighborsClassifierKNeighborsRegressor) make_pipeline)FunctionTransformerscale)SparseRandomProjection)SVCSVR)"ConsumingClassifierWithOnlyPredict)ConsumingClassifierWithoutPredictLogProba&ConsumingClassifierWithoutPredictProba _Registrycheck_recorded_metadata)DecisionTreeClassifierDecisionTreeRegressor)check_random_state)assert_allcloseassert_array_almost_equalassert_array_equal)CSC_CONTAINERSCSR_CONTAINERSctd}ttjtj|\}}}}t ddgddgddgddgd }dt td td ttg}t|t|D]3\}}td||d d|j||j|5y)Nr random_state??TF max_samples max_features bootstrapbootstrap_featuresmax_iter) max_depth) estimatorr2 n_estimators)r*ririsdatatargetrr rr(rr!ziprrfitpredict) rngX_trainX_testy_trainy_testgrid estimatorsparamsrAs a/var/www/auto_recruiter/arenv/lib/python3.12/site-packages/sklearn/ensemble/tests/test_bagging.pytest_classificationrSHs Q C'7 4;;S($GVWf :F#'-   D B+  J!uZ'890     #gw  0z sparse_container, params, methodr3r?Tr7r4r6Fr9r:r;r8r:r;)rI predict_probapredict_log_probadecision_functionclGddt}td}tttj tj |\}}}}||} ||} td |dddd |j| |} t| || } td |dddd |j||} t| ||}t| |t| }| jDcgc]}|j}}t|Dcgc]}||k( c}sJycc}wcc}w) Nc"eZdZdZfdZxZS)-test_sparse_classification..CustomSVC7SVC variant that records the nature of the training setcHt|||t||_|SNsuperrHtype data_type_selfXy __class__s rRrHz1test_sparse_classification..CustomSVC.fit! GK1 "1gDOKrT__name__ __module__ __qualname____doc__rH __classcell__rhs@rR CustomSVCr\E  rTrqrr1linearovr)kerneldecision_function_shaper5rAr2rC)r!r*rrrDrErFrrHgetattrr,rb estimators_rcall)sparse_containerrQmethodrqrJrKrLrMrNX_train_sparse X_test_sparsesparse_classifiersparse_resultsdense_classifier dense_results sparse_typeitypests rRtest_sparse_classificationrisK4C Q C'7 dii$++C($GVWf&g.N$V,M)8UK  c.'"  8W.7 FN)8UK  c'7  6G,f5f=Mnm<~&K#4#@#@ AaQ\\ AE A %0Q[ 0 11 1 B0s 8D, D1cttd}ttjddtjdd|\}}}}t ddgddgddgddgd}dt tttfD]6}|D]/}td ||d |j||j|18y) Nr2r1r3r4TFr7rwrC) r*rdiabetesrErFrr r)rr"rrHrI)rJrKrLrMrNrOrArQs rRtest_regressionrs Q C'7 crHOOCR0s($GVWf : #J#'-   D      F  Mys Mf M Q Q gfo  rTr{ctd}ttjddtjdd|\}}}}Gddt }ddddd d d ddd dd dd ddd dg}||}||} |D]} t d|dd| j||} | j| } t d|dd| j||j|} t|}| jDcgc]}|j}}t| | t|Dcgc]}||k( c}sJt| | ycc}wcc}w)Nrrr1c"eZdZdZfdZxZS))test_sparse_regression..CustomSVRr]cHt|||t||_|Sr_r`rds rRrHz-test_sparse_regression..CustomSVR.fitrirTrjrps@rR CustomSVRrrrrTrr3r?Tr7r4r6FrUrVr5rwrC)r*rrrErFr"rrHrIrbryrcr,rz)r{rJrKrLrMrNrparameter_setsr}r~rQrrrrrrrs rRtest_sparse_regressionrs Q C'7 crHOOCR0s($GVWfC"&  "&  dK$eLN"&g.N$V,M A, k 5; #ng & +22=A  My{ Mf M S' " WV_  >* '8'D'DE!EE!.-@e4A$4555!.-@'AF5s <E & E ceZdZdZdZy)DummySizeEstimatorc`|jd|_tj||_yNr)shapetraining_size_joblibhashtraining_hash_rerfrgs rRrHzDummySizeEstimator.fits"ggaj$kk!nrTcFtj|jdSr)nponesrrerfs rRrIzDummySizeEstimator.predictswwqwwqz""rTN)rkrlrmrHrIrCrTrRrrs -#rTrc td}ttjtj|\}}}}t j ||}tt dd|j ||}|j|||j||k(sJtt dd|j ||}|j|||j||kDsJttdj ||}g}|jD];}|j|jdk(sJ|j|j=tt!|t|k(sJy)Nrr1r4F)rAr8r:r2T)rAr:)r*rrrErFr)rHrscorerryrrappendrlenset)rJrKrLrMrNrAensemble training_hashs rRtest_bootstrap_samplesrsw Q C'7 xS($GVWf&'++GW=I ')   c'7   ??7G ,w0P PP P ')   c'7   ??7G ,x~~gw/O OO O  *<*>$OSSHM))7 ''7==+;;;;Y5567 s=! "c-&8 88 8rTc`td}ttjtj|\}}}}t t dd|j||}|jD]D}tjjdtj|jdk(rDJt t dd|j||}|jD]D}tjjdtj|jdkDrDJy)Nrr1r4F)rAr9r;r2r5T) r*rrrErFrr)rHestimators_features_rrunique)rJrKrLrMrNrfeaturess rRtest_bootstrap_featuresr1s Q C'7 xS($GVWf ')    c'7  11F}}""1%8)<)B)B1)EEEEF ')   c'7  11E}}""1% ((;(A(A!(DDDDErTc td}ttjtj|\}}}}t j dd5tt|j||}tt j|j|dt jt|t|j|t j|j!|tt#|d j||}tt j|j|dt jt|t|j|t j|j!|dddy#1swYyxYw) Nrr1ignore)divideinvalidrwr5)axis)rAr2r8)r*rrDrErFrerrstaterr(rHr,sumrWrrexprXrrJrKrLrMrNrs rRtest_probabilityrMsY Q C'7 4;;S($GVWf Hh 7 $,.S #gw   " FF8))&1 :BGGCK#C,,C5 ctd}ttjtj|\}}}}t t ddd|j||}|j||}t||jz dksJd}tjt|5t t d dd|}|j||dddy#1swYyxYw) Nrr1rTrrrrr5)r*rrrErFrr)rHrrrrrr) rJrKrLrMrNrrrregrs rRtest_oob_score_regressionrs Q C'7 xS($GVWf ')    c'7 66*J zCNN* +c 11 1 F  k 2#+-   '"###s /+C##C,cPtd}ttjtj|\}}}}t t ddd|j||}t j||}t|j||j|y)Nrr1r5F)rArBr:r;r2) r*rrrErFrrrHr,rI)rJrKrLrMrNclf1clf2s rRtest_single_estimatorrs Q C'7 xS($GVWf %'     c'7    $ $Wg 6Ddll62DLL4HIrTctjtj}}t}t t |j ||drJy)NrY)rDrErFr(hasattrrrH)rfrgbases rR test_errorrsB 99dkkqA ! #D(.221a8:MNN NNrTcttjtjd\}}}}t t ddj ||}|j|}|jd|j|}t||t t ddj ||}|j|}t||t tdddj ||}|j|}|jd|j|} t|| t tdddj ||}|j|} t|| y) Nrr1n_jobsr2r5rrt)rv) rrDrErFrr(rHrW set_paramsr,r!rY) rKrLrMrNry1y2y3 decisions1 decisions2 decisions3s rRtest_parallel_classificationrsj'7 4;;Q($GVWf!  c'7    'B q!    'Bb"%   c'7     'Bb"%! E*11 c'7 ++F3J q!++F3Jj*5  E*11 c'7 ++F3Jj*5rTctd}ttjtj|\}}}}t t ddj||}|jd|j|}|jd|j|}t||t t ddj||}|j|}t||y)Nrr1rrr5rr?) r*rrrErFrr)rHrrIr,) rJrKrLrMrNrrrrs rRtest_parallel_regressionrs Q C'7 xS($GVWf 5 7PQRVVH q!   & !B q!   & !Bb"% 5 7PQRVVH   & !Bb"%rTctjtj}}d||dk(<ddd}tt t |dj ||y)Nr5r?)r5r?)rB estimator__Croc_auc)scoring)rDrErFrrr!rH)rfrg parameterss rRtest_gridsearchrsP 99dkkqAAa1fI#)&AJ"35):yIMMaQRSrTctd}ttjtj|\}}}}t dddj ||}t|jtsJt tddj ||}t|jtsJt tddj ||}t|jtsJttjtj|\}}}}tdddj ||}t|jtsJttddj ||}t|jtsJttddj ||}t|jtsJy)Nrr1rr)r*rrDrErFrrH isinstance estimator_r(rrrr)r"rs rRtest_estimatorr(s Q C(8 4;;S($GVWf!aa@DDWgVH h))+A BB B   c'7  h))+A BB B aaHLLH h)): 66 6(8 xS($GVWf QQ?CCGWUH h))+@ AA A 5 7PQRVVH h))+@ AA Aaa@DDWgVH h))3 // /rTctttdtd}|j t j t jt|djddjtsJy)Nr5)kr?)r9r) rrrr(rHrDrErFrstepsr2intrAs rRtest_bagging_with_pipelinerUsd!kA&(>(@APQIMM$))T[[) il((,Q/<&) QUF   <  8 1a6{l***+"luIMM!Q f5d!!5 6#'01t  1;  51s C *C ctdd\}}tdd}|j|||jdt j t 5|j||dddy#1swYyxYw) Nr<r5rrT)rBrr6r)r rrHrrraises ValueErrorrfrgrs rR$test_warm_start_smaller_n_estimatorsrwsg bq 9DAq t D0D<$D-0D9 5D<<EceZdZdZddZdZy)EstimatorAcceptingSampleWeightz&Fake estimator accepting sample_weightNc.||_||_||_yzRecord values passed during fitN)X_y_sample_weight_)rerfrgrs rRrHz"EstimatorAcceptingSampleWeight.fits+rTcyr_rCrs rRrIz&EstimatorAcceptingSampleWeight.predict rTr_rkrlrmrnrHrIrCrTrRrrs0,  rTrceZdZdZdZdZy)EstimatorRejectingSampleWeightz&Fake estimator rejecting sample_weightc ||_||_yr)rrrs rRrHz"EstimatorRejectingSampleWeight.fitsrTcyr_rCrs rRrIz&EstimatorRejectingSampleWeight.predictr!rTNr"rCrTrRr$r$s0  rTr$ bagging_classaccept_sample_weightmetadata_routingr8rg?ctjdjdd}tjddgd}tjd}d|d<d|d<|r t }n t }|j\}} t|trt||jz} n|} t| 5|r|r|jd }|||d } | j||| t| j | j"D]P\} } tj$| | }t|t'| cxk(r| k(sJJtj(| ddgj+sJ|r|| j,j|| fk(sJ| j.j|fk(sJt1| j,|t1| j.|t1| j2|| j,j| | fk(sJ| j.j| fk(sJt1| j,|| t1| j.|| S dddy#1swYyxYw)Nrrr5rrr6r?renable_metadata_routingTr)r8rB) minlength)rarangereshaperepeatzerosrr$rrfloatrrrset_fit_requestrHrGryestimators_samples_bincountrisinrzrrr+r)r'r(r)r8rfrgrbase_estimatorr n_featuresexpected_integer_max_samplesbaggingrAsamplescountss rR%test_draw_indices_using_sample_weightr=s7 #r1%A 1a&"AHHSMMM!M!7979GGIz+u%(+;9J9J9L+L'M$'2$ 0@ A:  4+;;$;ONKVWX Aq 6"%g&9&97;V;V"W : Iw[[I>Fv;#g,N2NN NN NN777QF+//1 11# ||))i-DDDD ||))i\999 a0 a0 8 8&A!||)).JJ-WWWW ||)).J-LLLL aj9 aj9# : :::s >F*I22I;c\d\}}}t||||k(sJd\}}}t||||k(sJd\}}}t|||dk(sJd\}}}d}tjt|5t|||t ||zk(sJ dddd\}}}tjt|5t|||dk(sJ dddd}t j jd }d d |jd }}}tjt|5t|||t ||jzk(sJ dddt jd }tj5tjdd\}}}t||||k(sJdd|jd }}}t|||t ||jzk(sJ dddy#1swYrxYw#1swYBxYw#1swYxYw#1swYyxYw)N)rNnot_used)rrr?)rh㈵>Nr5)rgQ?Nz?.+the number of samples.+low number.+max_samples.+as an integerrzI.+the total sum of sample weights.+low number.+max_samples.+as an integerri@Br@)sizererror)rNrr3)rrrrrrrandom default_rnguniformrrwarningscatch_warnings simplefilter)rr8r warning_msgwarning_msg_with_weightsrJs rRtest_get_n_samples_bootstraprL3sj,@)I{M #I{M Ji WW W,=)I{M KG;VV-;)I{M #I{M Ja OO O,:)I{MSK k 5 ' ; NRU ) #S     -;)I{M k 5T' ; NRSSSST T ))   "C,5tS[[i[=XM{I k)A B ' ; NRU -++- -S     GGAJM  "  g&0=- ; $Y ] K   14S#++3+:O ; ' ; NRU -++- -S       )  TT    s1,G<5H #-HA0H"<H HH"H+c"tdd\}}tdd}|j|||jddd |j||t j t 5t|d dddy#1swYyxYw) Nrr5rrT)rBrFr)rrrBr)r rrHrrrAttributeErrorrxrs rR$test_oob_score_removed_on_warm_startrOdsu c :DAq d ;CGGAqMNNde"NEGGAqM ~ &#\"###s / BBctdd\}}ttdddd}|j||j|j||jk(sJy)Nr5rr3T)r8r9rr2)r rrrHrrfrgr:s rRtest_oob_score_consistencyrSqsb c :DAq G ;;q!  ' '7;;q!+<+G+G GG GrTctdd\}}ttdddd}|j|||j}|j }|j }t|t|k(sJt|dt|dzk(sJ|djjd k(sJd}||}||}||} ||dd|f} ||} | j} | j| | | j} t| | y) NrQr5rr3F)r8r9r2r:rr?r) r rrrHr4rryrdtypekindcoef_r,)rfrgr:estimators_samplesestimators_featuresrPestimator_indexestimator_samplesestimator_featuresrArKrM orig_coefs new_coefss rRtest_estimators_samplesr_s1 c :DAq G KK1!44!66$$J ! "c*o 55 5 !!$ %Q1 44 4 a & & + +s 22 2O*?;,_=?+I"#Q(:%:;G!"GJ MM'7#Ij)4rTct}|j|j}}tt dt }t |dd}|j|||jdjddjj}|jd}|jd}|jd}||dd|f} ||} |j| | t|jddj|y)Nr?) n_componentsr3r)rAr8r2rr5)rrErFrr rrrHryrrWcopyr4rr-) rDrfrg base_pipelinerpipeline_estimator_coefrAestimator_sampleestimator_featurerKrMs rR%test_estimators_samples_deterministicrgs ;D 99dkkqA!A.0B0DM mST UCGGAqM!ooa066r:1=CCHHJ"I..q1003!"A'8$89G !G MM'7#yr*1-335LMrTcd}td|zd\}}tt|dd}|j|||j|k(sJy)Nrr?r5rr3)r8r9r2)r rrrH _max_samples)r8rfrgr:s rRtest_max_samples_consistencyrjsZK a+oA FDAq G  KK1   ; .. .rTcHd}dgdgdggdz}gddz}gddz}gddz}td| j||j}td| j||j}td| j||j}||g||gk(sJy) Nrrrr5)ABC)rrr5)rr5r?T)rr2)rrHr)r2rfY1Y2Y3x1x2x3s rR!test_set_oob_score_label_encodingrusL sQC1A 1 B aB QBD|D Q  D|D Q  D|D Q  8Bx  rTc^|jdd}d|tj|<|S)Nr2T)rbr)astyperisfinite)rfs rRreplacerys- t$AAr{{1~o HrTc tjgdgddtjdgdtjdgdtj dgg}tjgdtjgdgdgdgdgdgg}|D]}t }t t t|}|j||j|t|}|j||j|}|j|jk(sJt }t |}tjt5|j||dddt|}tjt5|j||dddy#1swYLxYw#1swY1xYw)Nr5rrr?Nr?r})r?rrrr)r?r5 )rr})rarraynaninfr)rrryrHrIrrrrr)rfy_valuesrg regressorpipelinebagging_regressory_hats rR*test_bagging_regressor_with_missing_inputsrs    N N O   A !     H()+  !4W!=yI Q""1%,X6!%%a+33A6ww%++%%%*+  + ]]: &  LLA  ,X6 ]]: & (  ! !!Q ' ( ((   ( (s F+ F7+F4 7G c ttjgdgddtjdgdtjdgdtj dgg}tjgd}t }t t t|}|j||j|t|}|j|||j|}|j|jk(sJ|j||j|t }t |}tjt 5|j||dddt|}tjt 5|j||dddy#1swYIxYw#1swYyxYw)Nr{r|r?r})rr}r}r}r})rrrrr(rrryrHrIrrrXrWrrr)rfrg classifierrbagging_classifierrs rR+test_bagging_classifier_with_missing_inputsrsh    N N O   A !A')J09:FH LLAq!*841a  & &q )E 77ekk !! !((+$$Q'()JZ(H z " Q*84 z "%q!$%%%%sF"F."F+.F7ctjddgddgg}tjddg}ttdd}|j ||y)Nr5r?rr6rg333333?)r9r2)rrrrrHrRs rRtest_bagging_small_max_featuresr:sR 1a&1a&!"A !QA 2 43UVWG KK1rTcXtjj|}|jdd}tjd}Gddt }t |dd}|j||t|jdj|jdy)N r6ceZdZdZdZy)8test_bagging_get_estimators_indices..MyEstimatorz7An estimator which stores y indices information at fit.c||_yr_)_sample_indicesrs rRrHz.MyEstimator.fitPs #$D rTN)rkrlrmrnrHrCrTrR MyEstimatorrMs E %rTrr5r)rArBr2) rrD RandomStaterandnr.r)rrHr-ryrr4)global_random_seedrJrfrgrrs rR#test_bagging_get_estimators_indicesrDs ))   2 3C "aA " A%+% []QR SCGGAqMsq)993;R;RST;UVrTzbagging, expected_allow_nanr5r=cV|jjj|k(sJy)z*Check that bagging inherits allow_nan tag.N)__sklearn_tags__ input_tags allow_nan)r:expected_allow_nans rRtest_bagging_allow_nan_tagrYs(  # # % 0 0 : :>P PP PrTr+modelr)rArBc`|jtjtjy)zAMake sure that metadata routing works with non-default estimator.NrHrDrErFrs rR"test_bagging_with_metadata_routingrks IIdii%rTzsub_estimator, caller, calleerIrXrWctjddgddgddgg}gd}dgd}}t}||}d |zd z} t|| d d t | } | j ||t| |tjddgddgddgg||t |sJ|D]}t|||||y)aTest that metadata routing works in `BaggingClassifier` with dynamic selection of the sub-estimator's methods. Here we test only specific test cases, where sub-estimator methods are not present and are not tested with `ConsumingClassifier` (which possesses all the methods) in sklearn/tests/test_metaestimators_metadata_routing.py: `BaggingClassifier.predict()` dynamically routes to `predict` if the sub-estimator doesn't have `predict_proba` and `BaggingClassifier.predict_log_proba()` dynamically routes to `predict_proba` if the sub-estimator doesn't have `predict_log_proba`, or to `predict`, if it doesn't have it. rr?r5r6r})r5r?ra)registryset__requestT)rmetadatarr)rfrr)objr|parentrrN)rrr&rxrrHrr') sub_estimatorcallercalleerfrgrrrrAset_callee_requestr:s rR3test_metadata_routing_with_dynamic_method_selectionr|s0 1a&1a&1a&)*AA c38M{Hx0I&:5*GI)*M)4G KK1GGV ((QFQFQF+ ,# x== '   rTc`|jtjtjy)z^Make sure that we still can use an estimator that does not implement the metadata routing.Nrrs rR-test_bagging_without_support_metadata_routingrs IIdii%rT)*)~rnrrG itertoolsrrrnumpyrrsklearnr sklearn.basersklearn.datasetsrrr sklearn.dummyr r sklearn.ensembler r rrrrrrsklearn.ensemble._baggingrsklearn.feature_selectionrsklearn.linear_modelrrsklearn.model_selectionrrrsklearn.neighborsrrsklearn.pipelinersklearn.preprocessingrrsklearn.random_projectionr sklearn.svmr!r"%sklearn.tests.metadata_routing_commonr#r$r%r&r' sklearn.treer(r) sklearn.utilsr*sklearn.utils._testingr+r,r-sklearn.utils.fixesr.r/rJrD permutationrFrApermrErrSmark parametrizerrrrrrrrrrrr thread_unsaferrrrrrr r r rrrr$r=rLrOrSr_rgrjruryrrrrrrrrrCrTrRrs $  "&GG9   ?1?QQG*<< G, ?{ t{{''( IIdO kk$  ? x++, d# //$'0B& ' # !!&*   # !!&*   U$ Od% P O%.&2/.&2R8+^n-LM5AN5Ap##'9TE8  F"&J!#HJ(O&6V&&4 T)0)0XC44$&&3,7@  ]    ]  +;=N*OP/%?+eT];S 2/:3<@Q/:d. b # H%5PN6 / 4 &(R%@W*! 91E FM 7C DdK -/ 0%8 #% %( QQ- ,!<1  +;!  & .& # /IF 5   ,-@)L -# . # T (a8  #4!#DSTU  & &rT