o tBh@sddlZddlZddlmZddlZddlmZddlm Z ddlm Z ddlm Z ddl m Z ddlmZdd lmZd d Zejd d dgddZddZejd d dgddZejdejejejgejdejejejgddZejdejejejgddZejdejdddgdd Zejdejdddgd!d"Zd#d$Zd%d&Z ejjd'gd(gd)ge!gd*gd+gej!gd,gd-ge"d.ej!gd/d0ej#d1gge"d.ej!gd/d0e$d2d1gge"d.ej!gd3gd4ge"d.ej!gd5d0ej#dgge"d.ej!gd5d0e$d2dgge"d.ggd6d7d8d9Z%ejd d dgejd:d;dgd?d@Z&ejd:d;dd|d}gfd~gdfgdddgfggdd7ddZ7ddZ8ejjd'gd)gd(ge!gdgdgej!gd-gd,ge"d.ggdd7ddZ9ejjdhej!d0dJgge"d.j1ej!d0digge"d.j1gdjgej,fej!dDdBggdkd.j1ej!dDdlggdkd.j1gdmgej2fej!d0dJgge"d.j1ej!d0digge"d.j1e!gdjgej,fggdd7ddZ:ddZ;ddZejdd die?dfddie?dfddde?dfddDde@dfd d ie@dfgddZAddZBddZCddZDddZEddZFddZGejdej#de$d2gddZHejddgdggd=didfgd)gd(gdgd=ejIdJe"d.idfgd)gd(gdgd=gdidfgddZJejd=d`dbggdgddZKejjddgdâgd>dgd7ddƄZLejdeegddɄZMejdejdddBiddiddidBddМdlddМgejddgdԢggddքZNejdejd=d~d>dJggdd؄ZOejd=d0gdiggddڄZPejdejdddbiddiddiddiddidbddМdlddМgddZQejd=d>dJggddZRejd=d0gdiggddZSddZTejddbdDdМddligddZUddZVddZWddZXejdddZYddZZejdddDdgddZ[ejddBdbdgddZ\ejdddidfddidfddidfgddZ]ddZ^ejdgdejdgdddZ_ejdejdddgejdej#dgdd Z`d d Zaejd dd gejd d dgddZbejd d dgddZcejd d dgddZdejd d dgddZeddZfejdej#dgddZgejd d dgejdej#dgddZhejjdhej!d0ej#gge"d.j1ej!d0dJgge"d.j1ej!d0ej#dige"d.gej,fej!d0ej#gge"d.j1ej!d0dJgge"d.j1ej!d0ej#dige"d.gej,fej!dej#ggejd.j1ej!dYggejd.j1e!ddZej#ggejfggdd7d d!Ziejd"e!d#ej#dYggj1e!d$ej#d#ggj1e!dZggfe!gd%gj1e!gd&gj1e!ej#ggfej!d'ej#dJgge"d.j1e!d#ej#d$ggj1ej!digge"d.fej!gd(ge"d.j1e!gd)gj1ej!ej#gge"d.fgd*d+Zjd,d-Zkd.d/Zlejd0d1ddggej!d1ddggd2d.ej!d1ddggd3d.gejd4dcddggej!dcddggd2d.ej!dcddggd3d.gd5d6Zmd7d8Znd9d:Zod;d<Zpejd=dd?ZqdS(@Nsparse)NotFittedError)assert_array_equal)assert_allclose)_convert_container) is_scalar_nan) OneHotEncoder)OrdinalEncodercCstgdgdg}t}tdd}||}||}|jdks$J|jdks+Jt|s2Jt|r9Jt|gdgdgt||dS)NrrrFrr )?rrr)rrrrr) nparrayr fit_transformshaperissparsertoarray)X enc_sparse enc_denseX_trans_sparse X_trans_denser/var/www/html/riverr-enterprise-integrations-main/venv/lib/python3.10/site-packages/sklearn/preprocessing/tests/test_encoders.py!test_one_hot_encoder_sparse_denses   r!handle_unknownignoreinfrequent_if_existcCstgdgdgdg}tgdg}tdd}||tjtdd ||Wdn1s7wYt|d}|||}t || tgd gt ||td d}tjtd d||WddS1s|wYdS) N)rr r)rrr )rrr )rrerrorr"Found unknown categoriesmatch)rrrrrrr42zhandle_unknown should be one of) rrr fitpytestraises ValueError transformcopyrrrr"rX2oh X2_passedrrr #test_one_hot_encoder_handle_unknown(s&         "r6cCsbtdgdgg}tddgd}d}tjt|d||WddS1s*wYdS)Nab categorieszqThis OneHotEncoder instance is not fitted yet. Call 'fit' with appropriate arguments before using this estimator.r))rrr r-r.rr0rencmsgrrr test_one_hot_encoder_not_fittedEs "r>cCsxtgdd}tddgd}t|d}|||}t||tgdgdgt||dS)N)11111111223334444)r55555r@r')rrrrrrrr) rrreshaper r,r1rr0rr2rrr +test_one_hot_encoder_handle_unknown_stringsQs   rG output_dtype input_dtypecCstjddgg|dj}tjddgddgg|d}td|d}t|||t||||td|dd}t|||t||||dS)Nrrdtypeauto)r:rKF)r:rKr) rasarrayTr rrrr,r0)rIrHr X_expectedr4rrr test_one_hot_encoder_dtypeds rPcCstd}|ddgddgd}tjgdgdg|d }t|d }t|||t| | ||t|d d }t|||t| | ||dS) Npandasr7r8rr ABrrrrrrrrrJF)rKr) r- importorskip DataFramerrr rrrr,r0)rHpdX_dfrOr4rrr !test_one_hot_encoder_dtype_pandasss   r[zignore::FutureWarning:sklearn get_namesget_feature_namesget_feature_names_outcCst}gdgdgdgdg}||t||}|dkr(t|tjs(Jtgd||gd}t||gd}tgd|tj t d d t||d d gWddS1sbwYdS) N)Malergirlr r )Female)r`r )r_3boy r )r_[r`r]) x0_Femalex0_Malex1_1x1_41x1_51x1_91x2_boyx2_girlx3_1x3_2x3_12x3_21x4_3x4_10x4_30)onetwothreefourfive) one_Femaleone_Maletwo_1two_41two_51two_91 three_boy three_girlfour_1four_2four_12four_21five_3five_10five_30z!input_features should have lengthr)ryrz) r r,getattr isinstancerndarrayrr]r-r.r/)r\r<r feature_namesfeature_names2rrr "test_one_hot_encoder_feature_namess.  "rcCsdt}tjddggtdj}||t||}tddg|t||dgd}tdd g|dS) Nuc❤t1dat2rJu x0_c❤t1x0_dat2un👍me)input_featuresun👍me_c❤t1u n👍me_dat2)r rrobjectrNr,rr)r\r<rrrrr *test_one_hot_encoder_feature_names_unicodes  rcCstddggj}t}|jgdgd|dgdgks"J||jdks.J|jgdgd||jdksCJdS) Nrr )rrr r r9r:)r r%)rrr r r%r) rrrNr set_params get_paramsrrr)rr4rrr test_one_hot_encoder_set_paramssrcCsNtdd}||}tddd}||}t||t|s#J|S)NrLr9Fr:r)r rrrrisspmatrix_csr)rr<Xtr1Xtr2rrr check_categorical_onehots    rrdefr7abcr r)rcrr)rr r)r8rScat)r7rTrrJ)r8rrr7rnan)Nrr)r7r r)NrN)mixednumericrz mixed-nanzmixed-float-nanz mixed-Nonezmixed-None-nanzmixed-None-float-nan)idscCstt|dddgf}t|ddgddggtt|ddddgf}t|gdgdgtdd|}t|gdgdgdS) Nrr)rrrrrrrrrLr9)rrrrr)rrrrr)rrrrr rr)rXtrrrr test_one_hot_encoders rsparse_FTdropfirstcCsgdgdgdg}t||d}||}tj|td}t|||ddgddgd dgg}t|d |d }||}t|}t||||durgdgdgdg}t||d d gddggdgd}||}tj|td}d|d<t|||ddgddgd dgg}t|ddgddgg|d}||}tj|td}d|d<d|dddf<t|||tgdgdg}td}t j t |d||WddS1swYdS)Nrr)rr rrrrJr rrr rL)rr:rrr)6r8)rr"r:)r rrr)rr:r"r rrrrr)Shape of the passed X data is not correctr)) r rrrrrinverse_transformreescaper-r.r/)r"rrrr<X_trexpr=rrr test_one_hot_encoder_inversesF        "rz X, X_transr rrrrrrryrzr{r8rrrrr)rrrrr)rrrrrcCs`t|d|}d}|rt|d}tjt|d||WddS1s)wYdS)zCheck that `inverse_transform` raise an error with unknown samples, no dropped feature, and `handle_unknow="error`. Non-regression test for: https://github.com/scikit-learn/scikit-learn/issues/14934 rzqSamples \[(\d )*\d\] can not be inverted when drop=None and handle_unknown='error' because they contain all zerosrr)N)r r,rr-r.r/r)rX_transrr<r=rrr ?test_one_hot_encoder_inverse_transform_raise_error_with_unknown>s  "rcCsJtjddgddgddggtd}tddd }||}t|||dS) Nr_rrar r rJ if_binaryFrr)rrrr rrr)roherrrr &test_one_hot_encoder_inverse_if_binary^s   r)rrN reset_dropcCstjddgddgddggtd}t|dd}||||}t||}|j|d t| ||t |||tt|||dS) Nr_rrar r rJFrr) rrrr r,r0rrrrr)r\rrrrrrrrr test_one_hot_encoder_drop_resetgs      rmethodr,r@@cCsLt}d}tjt|dt|||WddS1swYdS)N'Expected 2D array, got 1D array insteadr))r r-r.r/r)rrr4r=rrr test_X_is_not_1Dws "rcCsdtd}|gd}t}d}tjt|dt|||WddS1s+wYdS)NrQ)r r%rrr))r-rWSeriesr r.r/r)rrYrr4r=rrr test_X_is_not_1D_pandass "rzX, cat_exp, cat_dtyperrr rSrT)rrrstringz missing-floatzmissing-np.nan-objectzmissing-float-nan-objectcCs||dddfD]Q}tdd}||t|jtsJt|j|D]6\}}|}t|drHt|ds9J|dd|ddksGJn||ksPJt |j |sYJq#q dS)NrCrLr9) r r,r categories_listziptolistrr issubdtyperK)rcat_exp cat_dtypeXir<resrres_listrrr test_one_hot_encoder_categoriess#   rzX, X2, cats, cat_dtypedr7r8cint64r%rr r )Nr7z)r7r8r)r7Nrr)rrz object-stringzobject-string-nonezobject-string-nanzobject-None-and-nanzobject-nan-and-NonecCst|d}tgdgdg}t|||t|jdt|dks)J|jd t|dks8J|jdj |ksBJt|d}t j t dd ||Wdn1s^wYt||d}tgdgdg}t||||dS) Nr9rrrrrrrr(r)r:r")rrr)r rrrrrrr:rrrKr-r.r/r,r0)rr3catsrr"r<rrrr )test_one_hot_encoder_specified_categoriess :   rcCsdtjddggtdj}tgdgd}tgdgdg}t||||t| |||j d gdksBJt |j dj tjsOJtd d ggj}tgd gd}d }tjt|d  | |Wdn1sywYtd d tjggj}td tjd ggd}tjt|d | |WddS1swYdS)Nr7r8rJ)r8r7rr9rrrrr )r rr z%Unsorted categories are not supportedr))rrrrNr rr,r0rrrrrrKobject_r-r.r/r)rr<rr=rrr (test_one_hot_encoder_unsorted_categories s$  "rcCstjddgddggtdj}tgdgdgd}tgd gd g}t||||jd gdks;Jt |jdj tj sHJ|jd  gdksUJt |jd j tj sbJdS) Nr7r8rr rJr)rrr r9)rrrrrr)rrrrrrr) rrrrNr rrrrrrrKrrr<rrrr 7test_one_hot_encoder_specified_categories_mixed_columns#srcCsDtd}|ddgddgd}t|}t|gdgdgdS) NrQr7r8rr rRrUrV)r-rWrXrr)rYrZrrrr test_one_hot_encoder_pandas0s rzdrop, expected_namesx0_cx2_br)rx1_2r)rr r8x0_bx2_a)rbinarymanualcCsVgdgdg}t|d}||t||}|dkr$t|tjs$Jt||dS)N)rr r7)r8r r8rr])r r,rrrrr)r\rexpected_namesrrrrrr 'test_one_hot_encoder_feature_names_drop:s   rcCsddgddgddgg}tgdgdgdg}tdd g}td d d }||}t|j|t||d dgddgddgg}tddgddgddgg}td dg}td d d }||}t|j|t||dS)Nrcyesnori)rrrrrE)rrrrrrFrtruer7falserr)rrr rr drop_idx_r)rexpectedexpected_drop_idxrresultrrr *test_one_hot_encoder_drop_equals_if_binaryPs        r)rcr r)rrr)rrrcCsTt}tjgdgdgdd}t|||dtdd}t|||dS)NrrrrrrrrJfloat64)r rrrrastyperrrr test_ordinal_encoderhs   r)rrzobject-string-catcCst|d}tdgdgg}t|||t|jdt|dks%J|jdt|dks4J|jdj |ks>Jt|d}t j t dd| |WddS1s[wYdS)Nr9rrrr(r))r rrrrrr:rrrKr-r.r/r,)rr3rrr<rrrr )test_ordinal_encoder_specified_categoriesys   "rcCsgdgdg}t}||}tj|td}t|||tgdgdg}td}t j t |d||WddS1sGwYdS)NrrrJ)rrrr rUrr)) r rrrrrrrrr-r.r/)rr<rrr=rrr test_ordinal_encoder_inverses   "rcCstddd}tjddgddgdd ggtd }tjdd gd dgddggtd }||||}tjd dgddgddggdd }t||||}tjddgddgddggtd }t||dS)Nuse_encoded_valuer" unknown_valuer7xr8yrrrJxyblar rrr)r rrrr,r0rr)r<X_fitr X_trans_encr X_trans_invinv_exprrr +test_ordinal_encoder_handle_unknowns_strings         rrKcCstddd}tjddgddgdd gg|d }tjdd gd dgddgg|d }||||}tjddgddgd d ggdd }t||||}tjddgddgddggtd }t||dS)Nrrrr r rJrfrr)r rrr,r0rrr)rKr<r rrrrrrrr ,test_ordinal_encoder_handle_unknowns_numerics         rzparams, err_type, err_msgrzbunknown_value should be an integer or np.nan when handle_unknown is 'use_encoded_value', got None.rrzTunknown_value should only be set when handle_unknown is 'use_encoded_value', got -2.r rzaunknown_value should be an integer or np.nan when handle_unknown is 'use_encoded_value', got bla.zhThe used value for unknown_value (1) is one of the values already used for encoding the seen categories.zKhandle_unknown should be either 'error' or 'use_encoded_value', got ignore.cCsftjddgddggtd}tdi|}tj||d||WddS1s,wYdS)Nr7r r8r rJr)r)rrrr r-r.r,)paramserr_typeerr_msgrencoderrrr *test_ordinal_encoder_handle_unknowns_raises % "rcCs`tdtjd}tdgdgdgg}|||dgdgdgg}t|dgdgtjggdS)Nrrrr r r%r)r rrrr,r0r)r<r rrrr (test_ordinal_encoder_handle_unknowns_nans  rcCsdtdtjtd}tdgdgdgg}tjtdd||WddS1s+wYdS)Nr)r"rrKrr r z'dtype parameter should be a float dtyper)) r rrintrr-r.r/r,)r<r rrr 8test_ordinal_encoder_handle_unknowns_nan_non_float_dtypes "rcCsjtjgdgtdj}gd}t|d}d}tjt|d||WddS1s.wYdS)N)LowMediumHighr!r rJ)r r!r"r9z*Shape mismatch: if categories is an array,r)) rrrrNr r-r.r/r,)rrr<r=rrr +test_ordinal_encoder_raise_categories_shapes  "r#csxtddtjgdgdgdd}tjddgd d ggd dtjddgd d ggddtd d gddggtddgddggtjdd gd d ggddfD]!tfddtdDscJt|qLddgd d ggtfddtdDsJt|dd gd d ggtfddtdDsJt|dS)NrLr9)rrrr)rrrrrrJrr r r%rr7r8rrabcdrcsg|] }j|jjkqSrrrK.0irr<rr *sz'test_encoder_dtypes..cs"g|] }tj|jtjqSr)rrrrKintegerr)r<rr r-/s"cg|] }j|jdkqS)rr(r)r/rr r-4) r rrr,allrangerr0r)rrr,r test_encoder_dtypess&     r4cstd}tddtjgdgdgdd}|jdd gd d gd d gddd}|tfddtd Ds.r7r8rrrSrTr5cs g|] }j|j|kqSrr(r)X_typer<rr r-Js ) r-rWr rrrXr,r2r3rr0rrK)rYrrrr6r test_encoder_dtypes_pandas8s  "   r8cCs*t}ddgddgg}tj|j|dS)Nr_rrar )r rtestingassert_no_warningsr)r<rrrr test_one_hot_encoder_warningNsr; missing_valuec Csndddd|g}t|d}gdgddddd|gg}||}gdgd gd g}t|||j|us8Jd d t|j|jD}||}t j |t d }t |drt|dd|ddt |dskJt |dssJt|ddddf|ddddft|dddf|dddft |dsJt |dsJdSt||t||dS)Nrrfr rr)rrfr rr7)rrfrrr7)rrrrr)rrrrrrcSsg|]\}}||qSrr)r*rfeaturerrr r-bs z4test_one_hot_encoder_drop_manual..rJrC)rCrC) r rrrrrrrrrrrr) r< cats_to_dropr<rtransr dropped_cats X_inv_transX_arrayrrr test_one_hot_encoder_drop_manualTs2      *" rCzX_fit, params, err_msgr_rasecondz Wrong input for parameter `drop`rr ;)ghir rFz&The following categories were supposedcCsLtdi|}tjt|d||WddS1swYdS)Nr)rr r-r.r/r,)r rrr<rrr #test_one_hot_encoder_invalid_paramsys "rI)rr rbr7cCs^t|d}d}tjt|d|gdgdgdgWddS1s(wYdS)Nrz-`drop` should have length equal to the numberr)rrrErH)rr<rrrr test_invalid_drop_lengths "rJdensityrdenser7r r8rcCst|d}t||d}gdgdg}||||t|j|j|dkr/t|jdnt||j|jD]\}}}|t||ksFJq7t|jtj sPJ|jj t ksXJdS)Nrr)rrr7rMrr) r r,rrrrrrrrrKr)rKrohe_baseohe_testrdrop_catdrop_idxcat_listrrr test_categoriess     rSEncodercCsd|dvs JdS)N categoricalX_types) _get_tags)rTrrr "test_encoders_has_categorical_tagssrXkwargsmax_categories min_frequency g(\?r)rZr[rfr:rL)r7r8rrc Cstdgddgddgddgdgj}td|d d d ||}t|jgd gdgdgdgdgd gg}tddgddgddgddgddgg}||}t||dddgdgdD}| |}t||| } tddg| | } tddg| dS)zpTest that different parameters for combine 'a', 'c', and 'd' into the infrequent category works as expected.r7rr8rrrcrr r$F)r:r"r)r7rrerrcSg|]}|gqSrrr*colrrr r-z2test_ohe_infrequent_two_levels..infrequent_sklearnr%rx0_infrequent_sklearnNr rrrNr r,rinfrequent_categories_r0rrr]r^) rYr:X_trainrX_testrr expected_invX_invrrrr test_ohe_infrequent_two_levelss,2(    rjcCstdgddgddgddgdgj}td d d |d |}t|jd gtdgdgg}||}td gdgg|| }tdg|| }tdg|| |}tdgdgg|dS)z3Test two levels and dropping the frequent category.r7rr8rrrcrr r$Fr r"rrZrrrrcrbN) rrrNr r,rrr0rr]r^r)rrfrrgrr X_inverserrr ,test_ohe_infrequent_two_levels_drop_frequents 2    rmcCstdgddgddgddgdgj}td d d |d }d |dd}tjt|d||WddS1sAwYdS)z_Test two levels and dropping any infrequent category removes the whole infrequent category.r7rr8rrrcrr r$Fr rkUnable to drop category r( from feature 0 because it is infrequentr)NrrrNr r-r.r/r,rrfrr=rrr 5test_ohe_infrequent_two_levels_drop_infrequent_errorss2 "rrrgQ?g{Gz?rc Cstdgddgddgddgdgj}tdd d d ||}t|jddggdgdgdgdgd gg}tgd gdgdgdgdg}||}t||dgdgdgdgdgg}| |}t||| }tgd|| }tgd|dS)zkTest that different parameters for combing 'a', and 'd' into the infrequent category works as expected.r7rr8rrrcrr r$Fr"rr]rrrrrrb)rrrcNrrd) rYrfrrgrrrhrirrrr test_ohe_infrequent_three_levels s22(    rucCstdgddgddgddgdgj}td d d|d |}tdgdgdgg}td d gd d gd d gg|||jdd|d}tj t |d|dgdgg}Wdn1sfwYtd d gd d gg|dS)z5Test three levels and dropping the frequent category.r7rr8rrrcrr r$Frkrrr#r'r(r)r]N) rrrNr r,rr0rr-warns UserWarning)rrfrrgr=rrrr .test_ohe_infrequent_three_levels_drop_frequent:s2"rxcCstdgddgddgddgdgj}td d d|d }d |d d}tjt|d||WddS1sAwYdS)z7Test three levels and dropping the infrequent category.r7rr8rrrcrr r$Frkrnrror)Nrprqrrr 7test_ohe_infrequent_three_levels_drop_infrequent_errorsOs2 "rycCstdgddgddgddgdgj}td d dd |}t|jddggdgdgdgdgg}tgd gd gdgd g}||}t||dgg}d}t j t |d||WddS1sowYdS)zmTest that different parameters for combining 'a', and 'd' into the infrequent category works as expected.r7rr8rrrcrr r&F)r"rrZrrtrbadz.Found unknown categories \['bad'\] in column 0r)N) rrrNr r,rrer0rr-r.r/)rfrrgrrr=rrr (test_ohe_infrequent_handle_unknown_error\s2"   "r{cCstjdgddgdgtdj}tdgdgddd ||}dgd gd gd gdgg}td dgdd gdd gdd gdd gg}||}t||dddgg}dgd gg}|D]}|j|d|tdgd gg||qZdS)zG'a' is the only frequent category, all other categories are infrequent.r7rr]rirJrrr7r8Fr$r:rr"r8rrrrrrrNr) rrrrNr r,r0rr)rYrfrrgrrdropsrrrr 5test_ohe_infrequent_two_levels_user_cats_one_frequentts("(    rcCstjdgddgddgddgdgtd j}tgd gd d d d|}t|jgdgdgdgdgdgdgg}tddgddgddgddgddgg}||}t ||dddgdgdD}| |}t||dS)zFTest that the order of the categories provided by a user is respected.r7rr8rrrcrr rJr|Fr$r r:rr"rZ)rrr7r]rrcSr^rrr_rrr r-raz.rbr%N rrrrNr r,rrer0rrrfrrgrrrhrirrr (test_ohe_infrequent_two_levels_user_catss**(   rcCstjdgddgddgddgdgtd j}tgd gd d dd |}t|jddggdgdgdgdgdgg}tgdgdgdgdgdg}||}t ||dgdgdgdgdgg}| |}t||dS)zTest that the order of the categories provided by a user is respected. In this case 'c' is encoded as the first category and 'b' is encoded as the second one.r7rr8rrrcrr rJ)rrr8r7Fr$rr]rrtrrbNrrrrr *test_ohe_infrequent_three_levels_user_catss4*(   rcCsbtjgdgdf}tdddd}||ddgddgg}||}t|gd gd gd S) zaTest infrequent categories where feature 0 has infrequent categories, and feature 1 does not. rrr r r r r rr ) rrrrrrrrrr rF)rZrrrrrrrr)rrrrN)rc_r r,r0r)rrrgrrrr test_ohe_infrequent_mixeds   rc Cstjgdgdgdf}tdddd}||}t|jdd d gt|jd d d gt|jd d d D]}t||}tgd|q:gdgdgdgdgdgdgdgdgdg }t||gdgdg}| |}gdgdg}t||| |}tj gdgdgt d} t| |tdddd |}tjtdd  | |Wd n1swYgd!gd"g}| |}gd#gdg}t||| |}tj gd$gd%gt d} t| |d S)&z?Test infrequent categories with feature matrix with 3 features.r) rrrrrrcrrr) rrrrrrrrrrLr r$r:rZr"rrr rcN)r]r^)x0_0x0_3rcx1_0x1_5x1_infrequent_sklearnx2_0x2_1)rrrrrrrr)rrrrrrrr)rrrrrrrr)rrrrrrrr)rrrrrrrr)rrrrrrrr)rrrrrrrr)rrrrrrrr)r rr )r%rr )rrrrrrrr)rrrrrrrr)r rbN)rbrNrJr&r(r))rrr)r rcr)rrrrrrrr)rbrbr)r rbr)rrr rrrrerrr0rrrr,r-r.r/) rrrr\rrrg X_test_transrirhrrr 'test_ohe_infrequent_multiple_categoriessr         rc Cstd}|jgdgddddgd}tdd d d }||}t|jd d dgt|jdgdgdgdgdgdgdgdgdgdgdg }t|||jddgddgdddgd}gdgdg}| |}t||| |}t j ddgddggt d}t|||jddgddgdddgd}| |}gdgdg}t||| |}t j ddgddggt d}t||dS)zHTest infrequent categories with a pandas dataframe with multiple dtypes.rQ) r7frrrr7rr8r8) rr rrcrcrfrr r)strrrrcolumnsrLr r$rrr7r8r)rr rf)rrrrrr)rrrrrr)rrrrrr)rrrrrr)rrrrrrrrfrbrJrrN)r-rWrXr rrrrerr0rrrr) rYrrrrrgrrirhrrr .test_ohe_infrequent_multiple_categories_dtypes?sV         rrh)r[rZcCsptdgddgddgddgdgj}tdd d d |}|||dgg}t|d ggd S),All user provided categories are infrequent.r7rr8rrrcrr r$FrsrNr)rrrNr r,r0rrYrfrrrrr $test_ohe_infrequent_one_level_errorss 2 rcCsbtjdgdgtdj}td gdgddd||}|dgdgg}t|d gd ggd S) rr]r rJr|Fr$r}r7rNr)rrrrNr r,r0rrrrr 5test_ohe_infrequent_user_cats_unknown_training_errorssrzkwargs, error_msgz%max_categories must be greater than 1rCz)min_frequency must be an integer at leastg?cCstdgddgddgddgdgj}td d d i|}tjt|d ||WddS1s:wYdS) Nr7rr8rrrcrr r"r$r)rrp)rY error_msgrfrrrr ,test_ohe_infrequent_invalid_parameters_errors 2  "rcCsbtjddggtdj}t|}d}tjt|d | WddS1s*wYdS)NrdogrJz&get_feature_names is deprecated in 1.0r)) rrrrNr r,r-rv FutureWarningr]r;rrr 1test_one_hot_encoder_get_feature_names_deprecateds   "rzinput_dtype, category_dtype)OOOUUOUUUSSOSUSS array_type)rr dataframec Cstjdgdgg|d}tjddg|dg}t|dd|}tdgdgdgdgg||d}||}tddgddgddgddgg}t||t|d|} | |}tdgdgdgdgg}t||d S) a"Check that encoding work with object, unicode, and byte string dtypes. Non-regression test for: https://github.com/scikit-learn/scikit-learn/issues/15616 https://github.com/scikit-learn/scikit-learn/issues/15726 https://github.com/scikit-learn/scikit-learn/issues/19677 r8r7rJFrrrr9N) rrr r,rr0rr r) rIcategory_dtyperrr:rrgrroerrr test_encoders_string_categoriess  "  rcCsTtjdd|d|ggtdj}tddd|}t||}t|ddd |gdS) Nr7r8rJFr#rr"x0_arx0_)rrrrNr r,rr)r\r<rrnamesrrr )test_ohe_missing_values_get_feature_namess rcCsrtd}|jgdtjdddtjgtdddd gd }tgd gd gd gdg}t|}t||dS)NrQ)rrNrr rr%rJ)col1col2rrr)rrrrrrr)rrrrrrr)rrrrrrr)rrrrrrr) r-rWrXrrrfloatrr)rYdfexpected_df_transrrrr %test_ohe_missing_value_support_pandass  r pd_nan_typepd.NAznp.nanc Cstd}|dkr |jntj}|d|jdd|ddgddi}tgd gd gd gd gd g}td |d}| |}t ||t |j dksMJt |j dddgdt|j ddsgJdS)NrQrrrr7r8categoryrJ)rrrr)rrrr)rrrrrFrrrrCr)r-rWNArrrXrrr rrlenrrisnan)rr"rYpd_missing_valuerrrdf_transrrr 1test_ohe_missing_value_support_pandas_categoricals(   rcCsddgddgddgg}tdd|d}||}tgd gd gd g}t||d d gg}tgd g}d}tjt|d ||}Wdn1sPwYt||| |}t |tjddggt ddS)zZCheck drop='first' and handle_unknown='ignore'/'infrequent_if_exist' during transform.r7rr8r rrFrrr"rr)rrrrr tFound unknown categories in columns \[0, 1\] during transform. These unknown categories will be encoded as all zerosr)NrJ r rrrrr-rvrwr0rrrr"rrrrOrgwarn_msgrirrr /test_ohe_drop_first_handle_unknown_ignore_warnss(      rcCsddgddgddgg}tdd|d}||}tgd gd gd g}t||d d gg}tgdg}d}tjt|d ||}Wdn1sPwYt||| |}t |tjddggt ddS)zDCheck drop='if_binary' and handle_unknown='ignore' during transform.r7rr8r rrFrrrrUrr )rrrrrr)NrJrrrrr 3test_ohe_drop_if_binary_handle_unknown_ignore_warns=s(      rcCsddgddgddgg}tdd|ddgddggd}||d dgg}tddgg}d }tjt|d  ||}Wd n1sDwYt||d S) znCheck drop='first' and handle_unknown='ignore'/'infrequent_if_exist' during fit with categories passed in.r7rr8r rrF)rrr"r:rzqFound unknown categories in columns \[0\] during transform. These unknown categories will be encoded as all zerosr)N) r r,rrr-rvrwr0r)r"rrrgrOrrrrr 'test_ohe_drop_first_explicit_categories`s    rcCsnttjdddggj}ttjd}dtj}tjt|d| |WddS1s0wYdS)zDTest ordinal encoder with nan passthrough fails when dtype=np.int32.rrrJzdThere are missing values in features \[0\]. For OrdinalEncoder to encode missing values with dtype: r)N) rrrrNr int32r-r.r/r,)rrr=rrr Btest_ordinal_encoder_passthrough_missing_values_float_errors_dtype{s  "rencoded_missing_valuecCstjtjdddggtjdj}t|d|}t|jdks Jt |jdddtjg| |}t ||gdgdgdgg| |}t ||dS) z.Test ordinal encoder with nan on float dtypes.rrrJrrrrN) rrrrrNr r,rrrr0r)rrrrrlrrr 5test_ordinal_encoder_passthrough_missing_values_floats  rc Cs td}|dkr |jntj}|d|jdd|ddgddi}t|d |}t |j d ks1Jt |j d d d gdt |j d dsKJ| |}t|dgdg|gdgdgg||}|jdkskJt |d dd fddgt |d d d fddgt |dsJd S)z0Check ordinal encoder is compatible with pandas.rQrrrr7r8rrJrrrNr rrC@rr)rrr r)r-rWrrrrXrr r,rrrrr0rrr)rrrYrrrrrlrrr =test_ordinal_encoder_missing_value_support_pandas_categoricals"   rr)zobject-None-missing-valuezobject-nan-missing_valueznumeric-missing-valuecCst|d}tdgtjgg}t||||jdj|ks!Jt|d}tj t dd| |WddS1s>wYdS)z.Test ordinal encoder for specified categories.r9rrr(r)N) r rrrrrrrKr-r.r/r,)rr3rrrrrrr =test_ordinal_encoder_specified_categories_missing_passthroughs &  "rzX, expected_X_trans, X_testrr)rrr)rrrr)rr7r8)rrrcCs8tddd}||}t||t||dggdS)z>Test the interaction between missing values and handle_unknownrrCrgN)r rrr0)rexpected_X_transrgrrrrr /test_ordinal_encoder_handle_missing_and_unknowns   rcCstgdgdg}t|}t}d}tjt|d ||Wdn1s,wYtjt|d | |Wdn1sHwY| |}t|}tjt|d| |WddS1sowYdS)zCheck that we raise proper error with sparse input in OrdinalEncoder. Non-regression test for: https://github.com/scikit-learn/scikit-learn/issues/19878 r rz6A sparse matrix was passed, but dense data is requiredr)N) rrr csr_matrixr r-r. TypeErrorr,rr)rX_sparserrrrrrr test_ordinal_encoder_sparses      "rcCstgdddtjf}tgdgddd}||tgdgdd}tjtd d ||WddS1s>wYdS) zCheck OrdinalEncoder.fit works with unseen category when `handle_unknown="use_encoded_value"`. Non-regression test for: https://github.com/scikit-learn/scikit-learn/issues/19872 )rrrrr rN)rCrrrr)r:r"rr&rr(r))rrnewaxisr r,r-r.r/)rrrrr -test_ordinal_encoder_fit_with_unseen_category*s   "rrfAAOUrgcCs4tddd}||||}t|ddggdS)zChecks that `OrdinalEncoder` transforms string dtypes. Non-regression test for: https://github.com/scikit-learn/scikit-learn/issues/19872 rirrN)r r,r0r)rfrgr<rrrr 1test_ordinal_encoder_handle_unknown_string_dtypes;s   rcCsbtgddd}t|}t|jtj|ddj| |}t|dgdgdgdggdS) zCheck that `OrdinalEncoder` accepts Python integers that are potentially larger than 64 bits. Non-regression test for: https://github.com/scikit-learn/scikit-learn/issues/20721 )l HP 1&l H]viel :?i}GalIRK2e6krCrr)axisr r N) rrrFr r,rrsortrNr0)rrrrrr #test_ordinal_encoder_python_integerWs  rcCsHtd}gd}|jgdg|d}t|}|}t||dS)z-Check feature names out is same as the input.rQ)r8rr7rrN)r-rWrXr r,r^r)rYrrr<feature_names_outrrr .test_ordinal_encoder_features_names_out_pandasks  rcCstjdgdgtjggtd}tdtjdd|}||}t|dgdgdggtjd gtjggtd}||}t|tjgdggd S) zECheck interactions between encode_unknown and missing value encoding.r7r8rJr)r"rrrrrN)rrrrr r,r0r)rrrrgrrrr 0test_ordinal_encoder_unknown_missing_interactionws  r with_pandascCstjddgddgdtjggtd}d}|r(td}|j|d d gd }|d }n|d }tdd}tjt |d| |WddS1sIwYdS)zXCheck OrdinalEncoder errors when encoded_missing_value is used by an known category.r7rr8rrrJzTencoded_missing_value \(1\) is already used to encode a known category in features: rQletterpetrz \['pet'\]z\[1\]rrr)N) rrrrr-rWrXr r.r/r,)rrrrYrrrr 0test_ordinal_encoder_encoded_missing_value_errors"    "r)rrnumpyrscipyrr-sklearn.exceptionsrsklearn.utils._testingrrr sklearn.utilsrsklearn.preprocessingr r r!mark parametrizer6r>rGrfloat32rrPr[filterwarningsrrrrrrrrrrrrrrrrr.str_float_rrNrrrrrrrrrrrrrrr/rrrr#r4r8r;rCrMrIrJrSrXrjrmrrrurxryr{rrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr s               A       -*       & &*  ! , 8         #     $             "    $  \A          #"   "