0Ph$QddlZddlZddlZddlZddlmZmZmZddl m Z m Z ddl m Z ddlmZddlmZddlmZddlmZdd lmZdd lmZmZmZdd lmZdd lmZd Z dZ!dZ"dZ#ej$%dedZ&ej$%dgdej$%dddgdZ'dZ(dZ)dZ*dZ+ej$%dgddZ,dZ-dZ.d Z/d!Z0ej$%dgdd"Z1ej$%d#gd$d%Z2d&Z3d'Z4d(Z5d)Z6d*Z7d+Z8dS),N) load_iris make_blobs make_circles)PCA KernelPCA)NotFittedError) Perceptron) rbf_kernel) GridSearchCV)Pipeline)StandardScaler)assert_allcloseassert_array_almost_equalassert_array_equal)CSR_CONTAINERS)_check_psd_eigenvaluesctjd}|d}|d}d}dD] }ddd|fD]}t | }t d ||| }||}|||} ttj |tj | |j dksJ||} | j d |j d ksJ|r'| | } | j |j ksJ d S) zNominal test for all solvers and all known kernels + a custom one It tests - that fit_transform is equivalent to fit+transform - that the shapes of transforms and inverse transforms are correct rrc`|iksJtj||S)N)npminimumsum)xykwargss k/var/www/html/test/jupyter/venv/lib/python3.11/site-packages/sklearn/decomposition/tests/test_kernel_pca.py histogramz"test_kernel_pca..histogram#s-||||z!Q##%%%autodensearpack randomizedlinearrbfpolyr)kernel eigen_solverfit_inverse_transformN)rrandom RandomState random_samplecallabler fit_transformfit transformrabssizeshapeinverse_transform) rngX_fitX_predr!r,r+invkpcaX_fit_transformedX_fit_transformed2X_pred_transformedX_pred2s r test_kernel_pcarCs )   " "C   f % %E   v & &F&&& B55  : 5 5Fv&&&C&|SVD!% 2 25 9 9 !%%!:!:5!A!A  %())262D+E+E    %)Q...."&!7!7 %+A.2C2I!2LLLLL 5001CDD} 44445 555r"ctddd}d}tjt|5|t jddddddS#1swxYwYdS)zCheck that kPCA raises an error if the parameters are invalid Tests fitting inverse transform with a precomputed kernel raises a ValueError. T precomputed) n_componentsr-r+z6Cannot fit_inverse_transform with a precomputed kernelmatchN)rpytestraises ValueErrorr4rr/randn) estimatorerr_mss r "test_kernel_pca_invalid_parametersrPFs tMIFF z 0 0 0// bioob"--...//////////////////s4A11A58A5cdtjd}|dd}t ||}||}|}d|dddf<||}t||dS)zCheck robustness to mutations in the original training array Test that after fitting a kPCA model, it stays independent of any mutation of the values of the original data object by relying on an internal copy. rrE) random_stateiN) rr/r0randrr4r5copyr)stateXr> transformed1X_copy transformed2s r $test_kernel_pca_consistent_transformrZTs I ! !! $ $E 2rA % ( ( ( , ,Q / /D>>!$$L VVXXFAaaadG>>&))LlL99999r"c tjd}|dd}d}|D]}tjd}t dD]6}t d||}||d||ddf<7t|tj |dddfd dddS) zTest that Kernel PCA produces deterministic output Tests that the same inputs and random state produce the same output. rrE)r&r%)rr\r)rGr,rRN) rr/r0rSzerosrangerr3rtilereshape)r:rVr,solver transformed_Xir>s r $test_kernel_pca_deterministic_outputrdgs )   " "C RA&LXX)) r ; ;A!&sSSSD"&"4"4Q"7"7":M!QQQ$   rw}QT/BB'G'G'O'OPRTU'V'VWWWW XXr" csr_containerc tjd}||d}||d}dD]}dD]}t d||dd}||}|||}ttj |tj |||} | j d |j d ksJtj t5|| d d d n #1swxYwY d S) zTest that kPCA works on a sparse data input. Same test as ``test_kernel_pca except inverse_transform`` since it's not implemented for sparse matrices. rrr)r$r&r')r(r)r*rF)r+r,r-rRr.N)rr/r0r1rr3r4r5rr6r8rJrKrr9) rer:r;r<r,r+r>r?r@rAs r test_kernel_pca_sparsergxs )   " "C M#++F33 4 4E ]3,,V44 5 5F8;; / ; ;F)&+ D!% 2 25 9 9 !%%!:!:5!A!A  %())262D+E+E    "&!7!7 %+A.2C2I!2LLLLL ~.. ; ;&&'9::: ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;- ;;;s7EE E rar# n_featuresrrEc tjd}|d|f}|d|f}|dkrdnd}t tjt ||||tjt||dkr|nd ||d S) zTest that kPCA with linear kernel is equivalent to PCA for all solvers. KernelPCA with linear kernel should produce the same output as PCA. rrrr&r)r,r%full) svd_solverN) rr/r0r1rr6rr4r5r)rarhr:r;r<n_compss r test_kernel_pca_linear_kernelrns )   " "C   q*o . .E   : / /F X%%aa1G yv666::5AAKKFSSTT  f.?.?FFV L L L SZZ Yv    r"c<tjd}|d}|d}dD]O}dD]J}t ||}|||j}|d|fksJKPdS) zTest that `n_components` is correctly taken into account for projections For all solvers this tests that the output has the correct shape depending on the selected number of components. rrrr%r&r')r.rrrGr,rN)rr/r0r1rr4r5r8)r:r;r<r,cr>r8s r test_kernel_pca_n_componentsrss )   " "C   f % %E   v & &F9##  # #A!,GGGDHHUOO--f55;EQF?????  ###r"cjtjddgddgddgg}t}||}|jdksJtd}||}|jdksJtdd}||}|jdksJd S) zCheck that the ``remove_zero_eig`` parameter works correctly. Tests that the null-space (Zero) eigenvalues are removed when remove_zero_eig=True, whereas they are not by default. ?r.)rjrrrG)rjrT)rGremove_zero_eigN)rarrayrr3r8)rVr>Xts r test_remove_zero_eigrzs 9a.1a&1i.9::A ;;D  A  B 8v     ! $ $ $D  A  B 8v     !T : : :D  A  B 8v      r"c&tjddgddgg}tj5tjdt tjd5tddd }|| |}| |}ttj |tj |d d d n #1swxYwYd d d d S#1swxYwYd S) zNon-regression test for issue #12141 (PR #12143) This test checks that fit().transform() returns the same result as fit_transform() in case of non-removed zero eigenvalue. r.rerrorwarn)allrFr%)rGrwr,N) rrxwarningscatch_warnings simplefilterRuntimeWarningerrstaterr4r5r3rr6)r;kABs r test_leave_zero_eigrst Hq!fq!f% & &E  " " < < g~666 [V $ $ $ < <q%gVVVAe &&u--A&&A %bfQii ; ; ; < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < < Kckpca_cs r )test_kernel_pca_precomputed_non_symmetricrs Q!RA 6PQ   D HHQKKK b'B7 B 6PQF JJrNNNt)6+?@@@t(&*=>>>>>r"cPtdddd\}}tdd}td |fd td fg}t d t jddz}t|d|}||||j dksJdS)zCheck that kPCA works as expected in a grid search pipeline Test if we can do a grid-search to find parameters to separate circles with a perceptron model. 333333?皙?r n_samplesfactornoiserRr)rr+rG kernel_pcar rmax_iter@)kernel_pca__gammarjcv param_gridr.N) rrr r dictraranger r4 best_score_)rVrr>pipeliner grid_searchs r test_gridsearch_pipeliner:s #cA N N NDAq E 2 2 2D,- jRS>T>T>T/UVWWHryQ/?/?(?@@@JxA*EEEKOOAq  "a ' ' ' ' ' 'r"cltdddd\}}tdd}td |fd td fg}t t jd d }t|d|}t|d}| |||j d ksJdS)zCheck that kPCA works as expected in a grid search pipeline (2) Test if we can do a grid-search to find parameters to separate circles with a perceptron model. This test uses a precomputed kernel. rrrrrrFrrrr rrr.)Perceptron__max_iterrjrr)gammaN) rrr r rrrr r r4r)rVrr>rrrX_kernels r $test_gridsearch_pipeline_precomputedrIs #cA N N NDAq M : : :D,- jRS>T>T>T/UVWWH29Q??;;;JxA*EEEK!3'''HOOHa    "a ' ' ' ' ' 'r"ctdddd\}}td||||}|dksJt d d d d }||}td||||}|dksJdS)aCheck that kPCA projects in a space where nested circles are separable Tests that 2D nested circles become separable with a perceptron when projected in the first 2 kPCA using an RBF kernel, while raw samples are not directly separable in the original space. rrrrrrrg?r)rTr)r+rGr-rruN)rr r4scorerr3)rVr train_scorer>rs r test_nested_circlesrYs #cA N N NDAqa(((,,Q2288A>>K      1D   D   " "Fa(((,,VQ77==faHHK #      r"c ddgddgddgg}tddd }|||jdksJt j|jt |jksJd S) z}Check that ``_check_psd_eigenvalues`` is correctly called in kPCA Non-regression test for issue #12140 (PR #12145). rr.gw̫@g:0yE>rr(rT)r+rGr-N)rr4rminrr~rrVr>s r test_kernel_conditioningrus Q(D!Ha=1A H1D Q Q QDHHQKKK   " "a ' ' ' ' 6$#'=d>O'P'PP Q QQQQQQr"c gdgdgdgdgdgdgdgdg}td |d }tjtd 5||dddn #1swxYwYtd |d }|dkrKtjtd 5||ddddS#1swxYwYdS||dS)aCheck how KernelPCA works with non-PSD kernels depending on n_components Tests for all methods what happens with a non PSD gram matrix (this can happen in an isomap scenario, or with custom kernel functions, or maybe with ill-posed datasets). When ``n_component`` is large enough to capture a negative eigenvalue, an error should be raised. Otherwise, KernelPCA should run without error since the negative eigenvalues are not selected. )gQ@p= # @p= ף@rr ףp= (\))rgQ@ףp= rrQ(\@)rrg(\.@Q@rr333333&{G:7)rrr@333333 rRQ?)rrrrrrrr)rrrrrrrr)rrrrrrgq= ףp@#@)rrrrrrrg(\u5@rF)r+r,rGz*There are significant negative eigenvaluesrHNrr')rrJrKrLr4)rarr>s r test_precomputed_kernel_not_psdrs <;;<<<;;;:::::::::;;;=== A MQ O O OD z)U V V V   MQ O O OD ] J      HHQKKK                    s$A**A.1A.(C  CCrG)rrEr\cd\}}t||zddd\}}|d|ddf||dddf}}t|dd||}t|d d||}t t j|t j|t|d d||} t t j| t j|dS) zGCheck that 'dense' 'arpack' & 'randomized' solvers give similar results)idrrrrNr%rr&r')rrr4r5rrr6) rGn_trainn_testrV_r;r<ref_preda_predr_preds r #test_kernel_pca_solvers_equivalencers? !OGV V#S1   DAqhwhkNAghhkN6E ,W1EEE U 6    ,XAFFF U 6   bfVnnbfX.>.>??? ,\JJJ U 6   bfVnnbfX.>.>?????r"c8tddd^}}tdddd }||}||}tj||z tj|z d ksJd S) zTest if the reconstruction is a good approximation. Note that in general it is not possible to get an arbitrarily good reconstruction because of kernel centering that does not preserve all the information of the original data. rrrrrhrRr\r)TgMbP?)rGr+r-alpha皙?N)rrr3r9rlinalgnorm)rVrr>X_trans X_reconsts r 0test_kernel_pca_inverse_transform_reconstructionrs  C C CEA T   D  ##G&&w//I 9>>!i- ( (29>>!+<+< s r test_32_64_decomposition_shapers yyy)))41RU   DAq &&q))ALA ;;D   a &$*<*z5test_kernel_pca_feature_names_out..s:::AA:::r"N)rrr4get_feature_names_outrr^)rVrr>namess r !test_kernel_pca_feature_names_outrst  C C CEA ! $ $ $ ( ( + +D  & & ( (E::q:::EBBBBBr"ctjd}|d}d|ddd}d|jdz }t d dd i||}t d d|i||}|j|ksJ|j|ksJ|| |}|| |}t||d S) z]Check that gamma is set correctly when not provided. Non-regression test for #26280 rrrTr))rGrRr-r+r.rNr) rr/r0r1r8rr4gamma_r9r5r)r:rVrexpected_gammakpca1kpca2X1_reconX2_recons r %test_kernel_pca_inverse_correct_gammars )   " "C &!!A!% F^N  + +D +F + + / / 2 2E  5 5N 5f 5 5 9 9! < ) ) ) ) <> ) ) ) )&&uq'9'9::H&&uq'9'9::HHh'''''r"ctjdtdd\}}tjd5t dd|ddddS#1swxYwYdS) zCheck that KernelPCA works with pandas output when the solver is arpack. Non-regression test for: https://github.com/scikit-learn/scikit-learn/issues/27579 pandasT)as_frame return_X_y)transform_outputrr&rqN)rJ importorskiprsklearnconfig_contextrr3)rVrs r test_kernel_pca_pandas_outputr-s  !!! dt 4 4 4DAq   : : :JJqx888FFqIIIJJJJJJJJJJJJJJJJJJs%A00A47A4)9rnumpyrrJrsklearn.datasetsrrrsklearn.decompositionrrsklearn.exceptionsrsklearn.linear_modelr sklearn.metrics.pairwiser sklearn.model_selectionr sklearn.pipeliner sklearn.preprocessingr sklearn.utils._testingrrrsklearn.utils.fixesrsklearn.utils.validationrrCrPrZrdmark parametrizergrnrsrzrrrrrrrrrrrrrrrrr"r rs @@@@@@@@@@00000000------++++++//////000000%%%%%%000000 /.....;;;;;;+5+5+5\ / / /:::&XXX".99";";:9";J#L#L#LMM2w//0/NM0###$,<<<2#O#O#OL#L#L#LMM??NM?4 ( ( ( ( ( ( 8 R R R #L#L#LMM00NM0f55!@!@65!@HDDD""""YYY"CCC(((8 J J J J Jr"