-i> ,SSKrSSKJr SSKrSSKrSSKJr SSKJ r J r SSK J r SSK JrJrJrJr SSKJr SSKJrJrJrJrJrJr SS KJrJrJrJrJ r J!r!J"r" SS K#J$r$J%r%J&r& SS K'J(r(J)r)J*r*J+r+J,r, SS K-J.r. SS K/J0r0 SSK1J2r2J3r3J4r4J5r5J6r6J7r7 SSK8J9r9 SSK:J;r;Jr>J?r? SSK@JArAJBrB SSKCJDrDJErEJFrFJGrGJHrHJIrI /SQrJSrKSrL\ R"5rN\NR\NRsrQrR\R"\QRS5rU\RRS5rX\XR\U5 \USSrU\Q\U\R\UsrQrR\ R"5r[\[R\[Rsr\r]Sr^Sr_\R"SS/S9S5ra\RRS\J5\RRSSS /5S!55rd\RRS\J5\RRSSS /5S"55re\RRS\J5\RRSSS /5S#55rf\RRS\J5\RRSSS /5S$55rg\RRS\J5\RRSSS /5S%55rh\RRS\J5\RRSSS /5S&55ri\RRS\J5\RRSSS /5\RRS'S/\G-5\RRS(S)S*/5S+5555rjS,rkS-rlS.rmS/rnS0roS1rp\RRS2/S3Q5\RRS4\G5S555rq\RRS6/S7Q5\RRS8SS /5\RRS4\G5S9555rr\RRS6/S7Q5\RRS8SS /5\RRS4\G5S:555rsSS<jrt\RRS=S>\"/S?QS/\G-555\RRS@/SAQ5SB55ru\RRSCSDSE/5\RRSF\R/\G-5\RRSGSHSI/5\RRSSS /5\RRSJ/SKQ5SL55555rwSMrx\RRSCSDSE/5\RRSF\R/\G-5\RRSNSOSP/5\RRSQ/SRQ5SS5555ry\RRS'S/\G-5\RRST/SUQ5SV55rzSWr{SXr|\RRSY\"S SZ9\4\"S SZ9\4/5S[5r}\RRSY\"5\4\"5\4/5\RRS\SS]/5S^55r~S_rS`rSarSbr\RRScSSd\^/5\RRS\S\*"Se5/5\RRS'S/\G-5Sf555r\RRS\S\*"Se5/5\RRS'S/\G-5Sg55rShrSir\RRSj\6"5\5Sk9\RRSl\B\/\9Sk9\RRSm\"SDSn9/\9Sk9So555r\RRSp\7"S Sq95Sr5r\RRSp\"\255Ss5r\RRSt\{\|\\\\45\RRS4\G5Su55rSvr\RRSw\\45Sx5rSyr\RRScSSz\_/5S{5r\RRScSSd\^/5S|5r\RRS}\\/5S~5r\RRS\SS]/5\RRS}\\/5S55rSrSr\RRSSS]/S]S//5\RRS'\E\F-\G-\H-\I-5S55rSr\RRS}\\/5\RRSSS0\S4SS0\S4SS0\S4/5S55r\RRS}\\/5S5rSr\RGR9S5S5r\RRS/SQ5\RRSSS /5\RRS4\G5S555r\RRS/SQ5\RRS4\G5S55r\RRSSS /5\RRS4\G5S55r\RRSS S/5\RRSS\GRB"S5/5\RRS\GRD/\G-5\RRS/SQ5S5555r\RRS/SQ5S5rSr\RRS/SQ5\RRS\"S;55S55rSr\RRS\04\S\S04\S\S]04/5S5r\RRSSS/5\RRSSS /5\RRS(/SQ5S555r\RRSSS /5\RRS(/SQ5S55r\RRS/SQ5S5r\RRS(/SQ5S5r\RRS(/SQ5S5rSr\RRSS S/5\RRS'S/\G-5\RRSSS/5\RRS\JS/-5S5555r\RRSS S/5\RRSS S/5\RRSS;S/5S555rSrSr\RRS\\/5\ "SS9S55r\RRS\"5\4\"5\4/5\ "SS9S55rg)N)product)linalg)config_contextdatasets)clone)make_classificationmake_low_rank_matrixmake_multilabel_classificationmake_regression)ConvergenceWarning)LinearRegressionRidgeRidgeClassifierRidgeClassifierCVRidgeCVridge_regression)_check_gcv_mode _RidgeGCV_solve_cholesky_solve_cholesky_kernel _solve_lbfgs _solve_svd_X_CenterStackOp) get_scorer make_scorermean_squared_error) GridSearchCV GroupKFoldKFold LeaveOneOutcross_val_predict) minmax_scale)check_random_state)_NUMPY_NAMESPACE_NAMES_atol_for_type_convert_to_numpy_get_namespace_device_dtype_ids)yield_namespace_device_dtype_combinationsyield_namespaces)_get_check_estimator_ids)assert_allcloseassert_almost_equalassert_array_almost_equalassert_array_equalignore_warnings)_array_api_for_tests check_array_api_input_and_values) _IS_32BITCOO_CONTAINERSCSC_CONTAINERSCSR_CONTAINERSDOK_CONTAINERSLIL_CONTAINERS)svd sparse_cgcholeskylsqrsagsaga)r9r<)r9r:r;r<r=c 2[R"X:H5$N)npmeany_testy_predkwargss X/var/www/html/venv/lib/python3.13/site-packages/sklearn/linear_model/tests/test_ridge.py_accuracy_callablerHYs 776# $$c,X- S-R5$)N)rB)rDrEs rG_mean_squared_error_callablerL]s _ " ( ( **rIlongwide)paramscURS:XaSup#OSup#[X#5n[RR U5n[ X#XES9nSUSS2S4'[ R"U5upxn [R"US:5(deUSS2SU24USS2US24pU SU2SS24XS2SS24pURS:Xa/URS S US 9nXn-nXURX#- S 9S --- nOGURS S US 9nU R[R"SU- 5-U R-U-nSnU[R"U5-nSUS'[ R"URU-U-URU-5nXU-- nXU-- n[R RU5[R RU5:deXoUU4$)aDataset with OLS and Ridge solutions, well conditioned X. The construction is based on the SVD decomposition of X = U S V'. Parameters ---------- type : {"long", "wide"} If "long", then n_samples > n_features. If "wide", then n_features > n_samples. For "wide", we return the minimum norm solution w = X' (XX')^-1 y: min ||w||_2 subject to X w = y Returns ------- X : ndarray Last column of 1, i.e. intercept. y : ndarray coef_ols : ndarray of shape Minimum norm OLS solutions, i.e. min ||X w - y||_2_2 (with minimum ||w||_2 in case of ambiguity) Last coefficient is intercept. coef_ridge : ndarray of shape (5,) Ridge solution with alpha=1, i.e. min ||X w - y||_2_2 + ||w||_2^2. Last coefficient is intercept. rM) )rRrQ) n_samples n_featureseffective_rank random_stateNMbP? lowhighsizer_rKr)rXrX)paramminrArandom RandomStater rr8alluniformnormalTdiagidentitysolvenorm)global_random_seedrequestrSrTkrngXUsVtU1U2Vt1_coef_olsyalphad coef_ridgeR_OLSR_Ridges rGols_ridge_datasetras>}} % : %  I"A ))   2 3C1 AAaeHzz!}HA" 66!d(    q"1"uXqABx AY2q5 }};;3Rj;A L #**)"8*9Q> >> KKCbyK 9552771q5>)BDD014 E  J''AAfIaccAgk13373J H Ej. G 99>>% 299>>'#: :: : : %%rIsolver fit_interceptTFcUupEpgSn[USUUS;aSOSUS9n U[R"U5- n XTU-- n S[R"U S-5[R"U S-5- - n [ S0U D6n US S 2S S 24nU(aUS nO%XDRS S 9- nXUR5- nS nU R XE5 US S nU R [R"U5:Xde[U RU5 U RXE5[R"U 5:Xde[ S0U D6R XE[R"URS 5S 9n U R [R"U5:Xde[U RU5 U RXE5[R"U 5:XdeU RU:Xdeg )zTest that Ridge converges for all solvers to correct solution. We work with a simple constructed data set with known solution. ?Tr<r=V瞯<绽|=r{rrtolrVrWrKNrXraxis sample_weight)dictrArBsumrfit intercept_pytestapproxr+coef_scoreonesshapesolver_)rrrrmrqrzrxcoefr{rOres_null res_RidgeR2_Ridgemodel intercepts rGtest_ridge_regressionrs&MA! E .EE' F2771:~HH I266)Q,'"&&1*===H OFOE !SbS& AH A  L  IIaO 9D   v}}Y7 77 7EKK& ;;q  h 7 77 7 OFO  BGGAGGAJ4G  HE   v}}Y7 77 7EKK& ;;q  h 7 77 7 ==F "" "rIc2UupEpgURupSn [U S- UUUS;aSOSUS9n USS2SS24nS [R"XD4S S 9-n[RR U5[ XS - 5::deU(aUSn O%XDRS S 9- nXUR5- nS n U RXE5 USSnU R[R"U 5:Xde[U R[RXw4S S9 g)zTest that Ridge converges for all solvers to correct solution on hstacked data. We work with a simple constructed data set with known solution. Fit on [X] with alpha is the same as fit on [X, X]/2 with alpha/2. For long X, [X, X] is a singular matrix. rrKrrrrNrX?rWrr:0yE>atol)rrrA concatenater matrix_rankrbrBrrrrr+rr_ rrrrmrqrzrxrrSrTr{rrs rG test_ridge_regression_hstacked_Xrs&MA!GGI E ai#.EE'  E !SbS& A bnnaV!,,A 99  #s91n'E EE EH A  L  IIaO 9D   v}}Y7 77 7EKKtz!2>rIc*UupEpgURupSn [SU -UUUS;aSOSUS9n USS2SS24n[R"XD4S S 9n[RR U5[ X5::de[RXU4nU(aUSn O%XDRS S 9- nXUR5- nS n U RXE5 USSnU R[R"U 5:Xde[U RUS S 9 g) a6Test that Ridge converges for all solvers to correct solution on vstacked data. We work with a simple constructed data set with known solution. Fit on [X] with alpha is the same as fit on [X], [y] [X], [y] with 2 * alpha. For wide X, [X', X'] is a singular matrix. rrKrrrrNrXrrrr)rrrArrrrbrrBrrrrr+rrs rG test_ridge_regression_vstacked_Xrs&MA!GGI E %i#.EE'  E !SbS& A vA&A 99  #s9'A AA A ad AH A  L  IIaO 9D   v}}Y7 77 7EKKD1rIcNUupEpgURupSn [U UUUS;aSOSUS9n [S 0U D6n U(aUSS2SS24nUSn USSnOSn U RXE5 X:dU(d=U R[ R "U 5:Xde[U RU5 g[U RU5U5 [XF-U -U5 [RR[RU RU R45[RR[RX45:de[ R"SS 9 U R[ R "U 5:Xde[U RU5 g) a Test that unpenalized Ridge = OLS converges for all solvers to correct solution. We work with a simple constructed data set with known solution. Note: This checks the minimum norm solution for wide X, i.e. n_samples < n_features: min ||w||_2 subject to X w = y rrrrrNrX1Ridge does not provide the minimum norm solution.reasonr)rrrrrrrr+rpredictrArrlrxfail)rrrrmrqrzrrxrSrTr{rOrrs rG!test_ridge_regression_unpenalizedr(sq&MA$GGI E #.EE' F OFOE a"fIH CRy  IIaO ]6==#;;;; T*  a(!,9,a0yy~~beeE$4$4ekk$ABCbiinn EE)/ "G      OP6==#;;;; T*rIc$UupEpgURupSn [U UUUS;aSOSUS9n U(aUSS2SS24nUSn USSnOSn S[R"XD4S S 9-n[RR U5[ X5::deU RXE5 X:dU(djU R[R"U 5:XdeUS :Xa[R"5 [U R[RXf45 g[U RU5U5 [RR![RU RU R45[RR![RXU45:de[R""S S 9 U R[R"U 5:Xde[U R[RXf45 g)aFTest that unpenalized Ridge = OLS converges for all solvers to correct solution. We work with a simple constructed data set with known solution. OLS fit on [X] is the same as fit on [X, X]/2. For long X, [X, X] is a singular matrix and we check against the minimum norm solution: min ||w||_2 subject to min ||X w - y||_2 rrrrrNrXrrWrr:rr)rrrArrrrbrrrrskipr+rrrrlr rrrrmrqrzrrxrSrTr{rrs rG,test_ridge_regression_unpenalized_hstacked_Xr_s&MA$GGI E #.EE'  E a"fIH CRy  bnnaV!,,A 99  #s9'A AA A IIaO]6==#;;;; Z  KKM RUU4:%67  a(!,yy~~beeE$4$4ekk$ABCbiinn EE)4' (G      OP6==#;;;; RUU4:%67rIcUupEpgURupSn [U UUUS;aSOSUS9n U(aUSS2SS24nUSn USSnOSn [R"XD4SS9n[RR U5[ X5::de[RXU4nU RXE5 X:dU(d=U R[R"U 5:Xde[U RU5 g[U RU5U5 [RR[RU RU R45[RR[RX45:de[R "S S 9 U R[R"U 5:Xde[U RU5 g) alTest that unpenalized Ridge = OLS converges for all solvers to correct solution. We work with a simple constructed data set with known solution. OLS fit on [X] is the same as fit on [X], [y] [X], [y]. For wide X, [X', X'] is a singular matrix and we check against the minimum norm solution: min ||w||_2 subject to X w = y rrrrrNrXrrr)rrrArrrrbrrrrrr+rrrlrrs rG,test_ridge_regression_unpenalized_vstacked_Xrs&MA$GGI E #.EE'  E a"fIH CRy  vA&A 99  #s9'A AA A ad A IIaO]6==#;;;; T*  a(!,yy~~beeE$4$4ekk$ABCbiinn EE)/ "G      OP6==#;;;; T*rIsparse_containerr{r{Gz?c UbMU(a U[;a[R"5 O&U(dU[;a[R"5 UupgpURup[ R SSU S9n [UUUUS;aSOSSUS 9n USS2SS 24n[R"Xf4SS 9n[RXw4n[RU SU - 4U-n U(aU S nO%XfRSS 9- nXwR5- nSnUbU"U5nU RXgU S 9 U SS n U R[R"U5:Xde[U R U 5 g) zTest that Ridge with sample weights gives correct results. We use the following trick: ||y - Xw||_2 = (z - Aw)' W (z - Aw) for z=[y, y], A' = [X', X'] (vstacked), and W[:n/2] + W[n/2:] = 1, W=diag(W) NrrWr\rrr順)r{rrrmax_iterrVrXrr)SPARSE_SOLVERS_WITH_INTERCEPTrr SPARSE_SOLVERS_WITHOUT_INTERCEPTrrprfrrArrrBrrrr+r)rrrr{rrmrqrzrxrrSrTswrrs rG$test_ridge_regression_sample_weightsrsU$# V+HH KKM61Q#Q KKM%MA!GGI  3B #.EE'  E !SbS& A vA&A ad A r1r6z U "BH A  L # Q  IIa"I% 9D   v}}Y7 77 7EKK&rIc,[RSS5n[[US/S9n[R "[[R 5n[X S/S9n[R "[R U5R n[X5 g)NrXrWrr{) y_diabetesreshaper X_diabetesrAdotrhrr-)rzrK dual_coefcoef2s rGtest_primal_dual_relationshiprsh2q!A :q 7D z:<<(A&qD6:I FF:<< + - -Ed*rIc [RRS5nURS5nURSS5nSn[R "[ US9 [X!SSSSS S 9 SSS5 g!,(df  g=f) Nrr[z3sparse_cg did not converge after [0-9]+ iterations.matchrr9rW)r{rrrverbose)rArcrdrandnrwarnsr r)rprzrqwarning_messages rG&test_ridge_regression_convergence_failr sj ))   "C ! A !RALO ( @ KS4QR  A @ @s A66 Bc[RRS5nSupURX5nURU5nUSS2[R4n[R USU-4n[ 5nURX45 URRU4:XdeURRS:Xde[UR[R5(de[UR[5(deURX55 URRU4:XdeURRS:Xde[UR[R5(de[UR[R5(deURX65 URRSU4:XdeURRS:Xde[UR[R5(de[UR[R5(deg)Nrrr[rWrrWrK)rK)rArcrdrnewaxisc_rrrrr isinstancendarrayfloat)rprSrTrqrzY1Yridges rGtest_ridge_shapes_typers ))   "C!I )(A )A 1bjj= B aQhA GE IIaO ;;   -- -    ! !R '' ' ekk2:: . .. . e&& . .. . IIa ;;   -- -    ! !T )) ) ekk2:: . .. . e&& 3 33 3 IIaO ;;  J // /    ! !T )) ) ekk2:: . .. . e&& 3 33 3rIc[RRS5nSupURX5nURU5n[RUSU-4n[ 5nUR X45 URnUR X55 [URSU5 [URSUS-5 g)NrrrrW) rArcrdrrrrrr,)rprSrTrqrzrrrs rGtest_ridge_interceptr5s ))   "C!I )(A )A aqjA GE IIaO  I IIaO((+Y7((+Y_=rIc[RRS5nSupURU5nURX5n[ SSS9n[ SS9nUR XC5 UR XC5 [URUR5 UR XC5 UR XC5 [URUR5 g)Nr)rrRrFr{rr) rArcrdrrr rr,r)rprSrTrzrqrolss rGtest_ridge_vs_lstsqrGs ))   "C I )A )(A 5 1E  /C IIaOGGAM SYY/ IIaOGGAM SYY/rIc [RRS5nSupnURX5nURX5n[R"U5n[R "[ XeR5VVs/sH(upx[USS9RXH5RPM* snn5n SV s/sH&n [XjSS9RXE5RPM( n n U Hn [X5 M [USSS 9n S n[R"[US 9 U RXE5 SSS5 gs snnfs sn f!,(df  g=f) N*)r[rr:r{r)r8r9r;r:r<r=-q=)r{rrrXrzCNumber of targets and number of penalties do not correspond: 4 != 5r)rArcrdrarangearrayziprhrrrr-rraises ValueError)rprSrT n_targetsrqrz penaltiesr{target coef_choleskyrcoefs_indiv_pencoef_indiv_penrerr_msgs rGtest_ridge_individual_penaltiesr\s4 ))   #C'0$I9 )(A )'A )$IHH"%Y!4 !4  j 1 5 5a @ F F!4 MNMF I%8<$>4@P#QR&**1-x||A?&((,,Q/1BCrIr))r[rW) r)r)rKrK)rruniform_weightsc[RRS5nUR"U6nU(a$[R"UR S5nOUR SUS5n[R"U5n[R"USUS9nXG- USS2S4-nURUR5n U"XFSS2S4-5n [SS9n U RX5up[X}5 [X5 gNrrW)rweightsTr)rArcrdrrr chisquaresqrtaveragerrhr _compute_gramr+)rr rrprqrrX_mean X_centered true_gramX_sparsegcv computed_gram computed_means rGtest_compute_gramrs ))   "C 5A WWQWWQZ  ]]1eAh 'ggbkG ZZ2 .F*4 00Jz||,IQD!112H $ 'C##4X#G MF*I-rIc[RRS5nUR"U6nU(a$[R"UR S5nOUR SUS5n[R"U5n[R"USUS9nXG- USS2S4-nURRU5n U"XFSS2S4-5n [SS9n U RX5up[X}5 [X5 gr)rArcrdrrrrrrrhrr_compute_covariancer+)rr rrprqrrrrtrue_covariancerr computed_covrs rGtest_compute_covariancer!s ))   "C 5A WWQWWQZ  ]]1eAh 'ggbkG ZZ2 .F*4 00J ll&&z2OQD!112H $ 'C"%"9"9("LLF*O2rIrWc  [UUUUUUUSU S9 upnUS:Xa[R"U/5nX- n [RR U 5R SX,R 5S:nU R5nSX)'SUU'U URU5-n U (aFXR[R"U5S-U- 5- n [R"U5S-nUS:XaUSnU (aXU4$X4$)NT) rSrT n_informativerbiasnoiseshufflerrVrWrr) r rAasarrayrcrdbinomialrcopyrabs)rSrTproportion_nonzeror#rr$X_offsetr%r&rpositiverVrqrzcmask removed_Xs rG_make_sparse_offset_regressionr1s#  ! GA!Q JJsOMA l+44Q8JGGTWXX IAeHIdOq A UU266!9q=1$ %% FF1IMQ aD Qw 4KrIzsolver, sparse_containerc#B# UHupUbUS;dMX4v M g7f)N)r9ridgecvr).0rrs rG r5s3+ &V  #v1I'I #"+ s )r:r<r9r;r=r3z"n_samples,dtype,proportion_nonzero))rfloat32皙?)(r6r)rfloat64皙?c SnUS:aSOSn[SSUUUUS9up[U5n[SUS 9RX5n UR US S 9nU R US S 9n UbU"U5nUS :Xa [ U/S 9n O,UR S5(aSn Sn OSn Sn [UUU U US9n U RX5 [U RU RSSS9 [U RU RSSS9 g)Nrg?gI@g@@r[)r$rTr+r%rVrSr8)rr{F)r)r3alphasr<gHz>rr)r{rrrrVrYrrtol) r1r"rrastyper startswithr+rr)rr+rSdtyperrmr{r%rqrz svd_ridgerrrs rGtest_solver_consistencyrEs$ E&,D%E ) -'  DA QAU%044Q:I U#A U#A# Q  w'   U # #CHCH+   IIaOEKKt$GE$$i&:&:DQrIgcv_moder8eigen X_containerX_shape)rr)rrzy_shape, noise))rr)rrW>@)rrb@c Uupg[U5S:XaUSOSn[UUUSSUSS9upU RU5n /SQn [UUU S S 9n [UUU S 9n U R X5 U"U 5nU R X5 U R [ R"U R 5:Xde[U RU RS S 9 [U RU RS S 9 g)NrKrXrWrFrrSrTrrVr&r%r#rYr7r$@g@@neg_mean_squared_errorcvrr>scoring)rFrr>rYr@) rr1rrralpha_rrr+rr)rFrHrIy_shaperr%rSrTrrqrzr> loo_ridge gcv_ridgeX_gcvs rGtest_ridge_gcv_vs_ridge_loo_cvr]s$I"7|q0 aI ) DA 'A (F #( I #I MM! NE MM%   v}}Y-=-=> >> >IOOY__4@I(()*>*>TJrIc SnSupSn[UUUSSSSS9upE/SQn[US X`S 9n[S X`S 9nURXE5 URXE5 UR[R "UR5:Xd#S UR<S UR<35e[ URURSS9 [ URURSS9 g)Nexplained_variance)r[rrWrFrrPrQTrT)rr>rVzgcv_ridge.alpha_=z, loo_ridge.alpha_=rYrW) r1rrrXrrr+rr) rVrSrTrrqrzr>rZr[s rGtest_ridge_loo_cv_asym_scoringr`Rs"G!II ) DA)F DId6KI MM! MM!   v}}Y-=-=> > 9   1 0 023 >IOOY__4@I(()*>*>TJrIrTrrzy_shape, fit_intercept, noise))rJTr)rKTg4@)rMTrN)rMFrLc &/SQn[RRS5n[U5S:XaUSOSn[ SUUSSUS9upU R U5n S UR [U 55-n XR5- S-R[5n [R"[R"U RS5U 5n U R[5n XXp[U RSS 9nURXU S 9n[!UUS US 9nUR#X5 [%UR&US9nURXU S 9n[)UXUS9nURUR:waUR UR5nUU- S-n[R"U RS5Vs/sHn[R*"UU U:HSS9PM nn[R,"U5nU"U 5n[!USUUS9nUR#UXS9 [U5S:Xa0UR.SS2SS2UR1UR&54nO,UR.SS2UR1UR&54nUR&[2R4"UR&5:Xde[7UUSS9 [7UR8UR8SS9 [7UR:UR:SS9 gs snf)NrQrrKrXrWrF)rSrTrrVr&r%r)n_splits)groupsrS)r>rUrVrrrUrT)r>store_cv_resultsrFrrrYrW)rArcrdrr1rrrbrAintrepeatrrrrsplitrrrrXr!rr' cv_results_indexrrr+rr)rFrHrrTrYr%r>rprrqrzrindicesX_tiledy_tiledrUsplitskfold ridge_reg predictions kfold_errorsir\r[ gcv_errorss rGtest_ridge_gcv_sample_weightsrurs)F ))   "C"7|q0 aI )  DA 'A #a&))M"%6%6%881<DDSIMii !''!*-}=G!((/Mz1:W QWWQZ (B XXgwX 7F  (#  E  IIgELL FI XXgwX 7F#IwFKKGMM)!))'--8 k)a/L<>IIaggaj >> >rIc[RSn/nUc[SpCO U"[5SpC[US9nURU[5 UR nUR U5 [n[[SS9n[SUS9n U"U R5"U[5 U R [R"U5:XdeSn [U 5n[SUS9n U"U R5"U[5 U R [R"U5:Xde[S5n [SU S9n U RU[5 U R [R"U5:XdeUcOURU[[R"U5S 9 UR [R"U5:Xde[R "[[45R"nURX>5 UR%U5nURU[5 UR%U5n['[R "UU45R"US S 9 U$) NrTFr)greater_is_better)rrVc[X5*$r@)r)xrzs rGfunc_test_ridge_loo..funcs"1(((rIrSrh㈵>rW)rrrrrrXappendr/rrrrrrrArvstackrhrr+)rrSretrqr ridge_gcvrXfrV ridge_gcv2r ridge_gcv3scorer ridge_gcv4rY_predrEs rG_test_ridge_loors  #I C%t=+J7= 6IMM!Z   FJJv A,FGug>Jjnna$    f 5 55 5)$Gug>Jjnna$    f 5 55 50 1Fuf=JNN1j!    f 5 55 5 a27793E F6==#8888 :z*+--A MM!   q !F MM!Z   q !FBIIvv./116E JrIcFUc[O U"[5n[5nURU[5 UR U5 [ UR R5S:Xde[UR5[RLde[S5nURUS9 URU[5 UR U5 [ UR R5S:Xde[UR5[RLdeg)NrWrrd)rrrrrrrrtyperrAr9r set_params)rrqridge_cvrUs rG_test_ridge_cvr s&. 4DZ4PAyH LLJ Q x~~## $ )) ) ## $ 22 2 qB 2 LLJ Q x~~## $ )) ) ## $ 22 2rIzridge, make_dataset)rec^U"SSS9up#URX#5 [US5(aeg)NrrSrVri)rhasattr)r make_datasetrqrzs rG$test_ridge_gcv_cv_results_not_storedrs3 !" 5DA IIaOum,,, ,,rIrUrcU"SSS9up4URSUS9 URX45 [US5(de[UR[ 5(deg)NrrrF)rerU best_score_)rrrrrr)rrrUrqrzs rGtest_ridge_best_scorer)s\ !" 5DA e3 IIaO 5- ( (( ( e'' / // /rIc x[RRS5nSupnURX5n[R"USS2S/4[R "SU455[R"USS2S/4S[R "SU45-5-[R"USS2S/4S[R "SU45-5-URX5-nSnUR Vs/sH%n[US 9RXW5RPM' nn[US S 9RXT5n [XR5 [[U RS 9RXT5RU R5 [US S S 9RXT5n U RRU4:XdeU RRU4:XdeU R RU[#U5U4:Xde[SS S S 9RXT5n U RRU4:XdeU RRU4:XdeU R RXS4:Xde[US S S 9RXTSS2S45n [R$"U R5(de[R$"U R5(deU R RU[#U54:Xde[US SS9RXT5n [XR5 [[U RS 9RXT5RU R5 [U['5S S9n Sn [(R*"[,U S9 U RXT5 SSS5 [USS S9n [(R*"[,U S9 U RXT5 SSS5 gs snf!,(df  NR=f!,(df  g=f)Nr)rrrrrWg?rKrY)rWdr=T)r>alpha_per_targetr)r>rrer2)r>rrV)r>rUrz3cv!=None and alpha_per_target=True are incompatiblerr)rArcrdrrrrhrrrXr.r-rrrrririsscalarr rrr) rprSrTrrzrqr>roptimal_alphasrmsgs rG"test_ridge_cv_individual_penaltiesr7s ))   #C(0$I9 )'A qQCy"''1j/23 &&1qc6D277Az?#;; < = &&1qc6EBGGQ O$<< = > ))I * +FRSQTQTUQTvgV,00;BBQTNUft<@@FH~7 HOO$((.44hnn ftdSWW H ?? YL 00 0    % %) 55 5    % %)S[))L LL La$NRRSTXH ?? YL 00 0    % %) 55 5    % %))B BB BftdSWW QT7H ;;x ' '' ' ;;x++ , ,, ,    % %)S[)A AA AftTJNNqTH~7 HOO$((.44hnn fNH ?C z - Q .fTBH z - Q . -aVZ . - . -s?,P<P:P+ P(+ P9cUc[O U"[5n[SS9nURU[5 [R "UR U[5S5$)NFrr)rrrrrAroundr)rrqrs rG_test_ridge_diabetesr~sH&. 4DZ4PA  &E IIa 88EKK:. 22rIcUc[O U"[5n[R"[[45Rn[R Sn[ SS9nURX5 URR SU4:XdeURU5nURU[5 URU5n[[R"Xf45RUSS9 g)NrWFrrKrdecimal) rrArrrhrrrrrr-)rrqrrTrrrEs rG_test_multi_ridge_diabetesrs&. 4DZ4PA :z*+--A!!!$J  &E IIaO ;;  J // / ]]1 F IIa ]]1 Fbii(89;;VQOrIcb[R"[5RSn[RSnUc[O U"[5n[ 5[ 54HjnURU[5 URRX4:XdeURU5n[R"[U:H5S:aMje [S5n[ US9nURU[5 URU5n[R"[U:H5S:deg)NrrWgHzG?rrdg?) rAuniquey_irisrX_irisrrrrrrBr)r n_classesrTrqregrErUs rG_test_ridge_classifiersrs &!''*IaJ"*0@0HA!#4#67 6yy9"9999Qwwv'(4/// 8 qB r "CGGAv [[^F 776V# $ ++ +rIrVaccuracyrcUc[O U"[5n[U5(a [U5OUn[XBS9nUR U[ 5R U5 g)N)rVrU)rcallablerrrrr)rrVrUrqscoring_clfs rG"test_ridge_classifier_with_scoringrsO#*0@0HA'/'8'8{7#gH H 4CGGAvq!rIcRSnUc[O U"[5n[R"SSSS9n[U[ U5US9nUR U[ 5 UR[R"S5:XdeUR[R"US5:Xdeg) Ncg)NzG?rrCs rG _dummy_score:test_ridge_regression_custom_scoring.._dummy_scoresrIrKr)num)r>rVrUrr) rrAlogspacerrrrrrrrX)rrUrrqr>rs rG$test_ridge_regression_custom_scoringrs #*0@0HA [[QA &F 6;|3LQS TCGGAv ??fmmD1 11 1 ::vay1 11 1rIcUc[O U"[5n[SSS9nURU[5 UR U[5n[SSS9nURU[5 UR U[5nX5:deg)NrF)rrrY)rrrrr)rrqrrridge2score2s rG_test_tolerancersr&. 4DZ4PA d% 0E IIa KK: &E t5 1F JJq* \\!Z (F ??rIc[X#5n[RU5n[RU5nUR XcS9nUR XsS9n UR Xg5 UR n URn [SS9 [U5R X5n U R n U RS:XdeU RUR:Xde[[XS9U [U5S9 U RnURS:XdeURUR:Xde[[XS9U [U5S9 SSS5 g!,(df  g=f)NdeviceTarray_api_dispatch)rR)xprr)r0rrArr'rrrrrrrCr+r&r%)name estimatorarray_namespacer dtype_namer X_iris_np y_iris_np X_iris_xp y_iris_xpcoef_np intercept_np estimator_xpcoef_xp intercept_xps rGcheck_array_api_attributesrs3 o 6B j)I j)I 9 4I 9 4I MM)'ooG''L 4 0Y'++IA $$}}$$$}} /// g -  + $.. !!R'''!!Y__444 l 2  +  1 0 0s CE E"z#array_namespace, device, dtype_name)idscheckrrc@URRnU"XPX#US9 g)N)rr) __class____name__)rrrrrrs rGtest_ridge_array_api_compliancers!$    ' 'D $?jQrIr)include_numpy_namespacesc[USS9nUR[SS5nUR[SS5n[R SSR nUSS1- Hgn[ XUS:HS9nS URS US 3n[R"[US 9 [S S9 URX#5 SSS5 SSS5 Mi [ SS S9nSURS3n[R"[US 9 [S S9 URX#5 SSS5 SSS5 [ 5nSURS3n[R"[US 9 [S S9 URX#5 SSS5 SSS5 g!,(df  N=f!,(df  GMK=f!,(df  N=f!,(df  N=f!,(df  N]=f!,(df  g=f)Nrrrrrvr8lbfgsrr-z Array API dispatch to namespace z" only supports solver 'svd'. Got 'z'.rTrzYThe solvers that support positive fitting do not support Array API dispatch to namespace zc. Please either disable Array API dispatch, or use a numpy-like namespace, or set `positive=False`.z&Using Array API dispatch to namespace z with `solver='auto'` will result in using the solver 'svd'. The results may differ from those when using a Numpy array, because in that case the preferred solver would be cholesky. Set `solver='svd'` to suppress this warning.)r0r'rrr_parameter_constraintsoptionsrrrrrrr UserWarning)rrrravailable_solversrr expected_msgs rG6test_array_api_error_and_warnings_for_solver_parameterr s od ;B 6"1:&I 6"1:&I44X>qAII#vuo5V.?@.r{{m<""( -  ]]:\ :48 )/9; :6 $ /E ++-;;-8. . z 6 t 4 IIi +5 7 GE 0 >D D k 6 t 4 IIi +5 7 6-98; :5 4 7 65 4 7 6sl F'F 9F G F0G G#)G;G# F F F- 0 F> :G G G G## G1c[USS9nUR[SS5nUR[SS5n[ 5nSn[ R "5 [ R"SU[S9 [SS9 URX#5 SSS5 SSS5 [SS9 [ SSS 9RX#5 SSS5 g!,(df  NB=f!,(df  NK=f!,(df  g=f) NrrzkResults might be different than when Array API dispatch is disabled, or when a numpy-like namespace is usederror)messagecategoryTrrvr) r0r'rrrwarningscatch_warningsfilterwarningsrrr)rrrrrrs rG)test_array_api_numpy_namespace_no_warningr:s od ;B 6"1:&I 6"1:&I GE ;  " T t 4 IIi +5 # 4 0 Vd+// E 1 0 5 4 # " 1 0s0$CCC+C0 C C C-0 C> test_funccHU"S5nU"U5nUbUb [X#SS9 ggg)Nrr)r-)rr ret_dense ret_sparses rGtest_dense_sparserRs4$I=)J!7!)C"8rIc[R"SS/SS/SS/SS/SS//5n/SQn[SS9nURX5 [ UR SS//5[R"S /55 [S S 0S9nURX5 [ UR SS//5[R"S /55 [S S9nURX5 [ UR SS//5[R"S /55 [R"SS/SS/SS/SS//5n/S Qn[SS9nURX5 [S S9nURX5 [ UR5S:Xde[URUR5 [URUR5 g)Nr皙rrrWrWrWrXrX class_weightr:rWrYrXbalanced)rWrWrXrXrK) rArrrr.rrclasses_r-rr)rqrzrregas rGtest_class_weightsrhsy 4,q D$<#sc3ZPQAA t ,CGGAMs{{S$K=1288QC=A 5z 2CGGAMs{{S$K=1288RD>B z 2CGGAMs{{S$K=1288QC=A 4,q D$<#sDEAA t ,CGGAM  3DHHQN t}}  "" "cii4cnndoo>rIrcU"5nUR[R[R5 U"SS9nUR[R[R5 [ UR UR 5 [ R"[RR5nU[RS:H==S-ss'SSSS.nU"5nUR[R[RU5 U"US9nUR[R[R5 [ UR UR 5 U"5nUR[R[RUS-5 U"US9nUR[R[RU5 [ UR UR 5 g ) z5Check class_weights resemble sample_weights behavior.rrrWrrgY@)rrWrKrKN) ririsdatarr,rrArr)rreg1reg2rrs rG"test_class_weight_vs_sample_weightrs= 5DHHTYY $ J 'DHHTYY $ DJJ/GGDKK--.M$++"#s*#u-L 5DHHTYY ]3 L )DHHTYY $ DJJ/ 5DHHTYY ]A%56 L )DHHTYY ]3 DJJ/rIc4[R"SS/SS/SS/SS/SS//5n/SQn[S/SQS9nURX5 [S S 0/S QS9nURX5 [ UR S S //5[R"S/55 g)Nrrrrrr)rr7rW)rr>rWrY)rr7rWr[gɿrKrX)rArrrr.r)rqrzrs rGtest_class_weights_cvr s 4,q D$<#sc3ZPQAA n ECGGAM !UrUrerVr)rUrerVzcv!=None and store_cv_resultsr)rArcrdrrrrrrrirrrr) rVrprSrTrr>n_alphasrrrzrs rGtest_ridgecv_store_cv_resultsrs  ))   #CIJ )(A F6{H'/'8'8{7#gHv$xPA )AEE!K ==  9"7 77 7I )'AEE!K ==  9"B BB B1tW=A z)H I a  J I Is D-- D;cD[R"SS/SS/SS/SS/SS//5n[R"/SQ5nURSn/SQn[U5n[ U5(a [ U5OUn[ USSUS 9nS nURX5 URRX8U4:Xde[R"/SQ/S Q/S Q/5R5nURS nURX5 URRX8U4:Xdeg) Nrrrrrrr Tr rW)rWrXrWrXrW)rXrXrWrXrX) rArrrrrrrri transpose) rVrrzrSr>r rrrs rG)test_ridge_classifier_cv_store_cv_resultsrs 4,q D$<#sc3ZPQA "#A I F6{H'/'8'8{7#gH$x A IEE!K ==  9"B BB B  -/BC ik IEE!K ==  9"B BB BrI Estimatorc[RRS5nSnSup4U[LaUR U5nOUR SSU5nUR X45nU"US9nUR ULdSURS35eURXe5 [UR [R"U55 g)Nrr rrrKr=z`alphas` was mutated in `z .__init__`) rArcrdrrrandintr>rrr.r')rrpr>rSrTrzrq ridge_ests rGtest_ridgecv_alphas_conversionrs ))   "C F IG IIi  KK1i ( )(A(I   v % #I$6$6#7zB %MM!y''F);rUNz"alphas\[0\] == 0.0, must be > 0.0.r) rArcrdrrrrrrr) rUrrpr>rSrTrzrqrs rGtest_ridgecv_alphas_zeror s ))   "C F IG IIi  KK1i ( )(A/I z ]]:-R S MM! T S  aT Ss B44 Cc[RRS5nSnSHup#URU5nURX#5nSUR U5-n[ S5n[ XS9nURXTUS9 SU0n [[5XS 9n U RXTUS9 URU RR:Xde[URU RR5 M g) Nrr )rrrrrrrr{rd)rArcrdrrandrrrrrrXbest_estimator_r{r-r) rpr>rSrTrzrqrrUr3 parametersgss rGtest_ridgecv_sample_weightr"s ))   "C F"3 IIi  IIi ,chhy11 1X/ A 6v& %': 5 q=1~~!3!3!9!9999!'--1C1C1I1IJ"3rIc^ ^ ^ ^^SS/nSS/n[RRS5n[X5GHup4UR X45m UR U5mUR U5S-S-nSnSnUSS2[R 4m U[R SS24m[ SS9m T RT TU5 T RT TU5 T RT TU5 U U U U4SjnU U UU4S jn S n [R"[U S 9 U"5 SSS5 S n [R"[U S 9 U "5 SSS5 GM g!,(df  N?=f!,(df  GMB=f) NrKrrrWrg@rc,>TRTTT5 gr@r)rqrsample_weights_not_OKrzsrGfit_ridge_not_okStest_raises_value_error_if_sample_weights_greater_than_1d..fit_ridge_not_okOs IIa1 2rIc,>TRTTT5 gr@r%)rqrsample_weights_not_OK_2rzsrGfit_ridge_not_ok_2Utest_raises_value_error_if_sample_weights_greater_than_1d..fit_ridge_not_ok_2Rs IIa3 4rIz)Sample weights must be 1D array or scalarr) rArcrdrrrrrrrr) n_sampless n_featuressrprSrTsample_weights_OKsample_weights_OK_1sample_weights_OK_2r'r+rrqrr&r*rzs @@@@@rG9test_raises_value_error_if_sample_weights_greater_than_1dr27sKQJa&K ))   #C!$Z!= IIi , IIi IIi0A59!! 1!RZZ- @"3BJJM"BA  !Q)* !Q+, !Q+, 3 3 5 5> ]]:W 5  6> ]]:W 5  6 57">.6 56 5sEE* E' * E: zn_samples,n_featuresrKcj[RRS5n[SSS9n[SSS9nUR X5nUR U5nUR U5S-S-nU"U5n UR XUS9 UR XgUS9 [ URURSS 9 g) NrrFrrKrWrrr)rArcrdrrrr-r) rSrTrrp sparse_ridge dense_ridgerqrzsample_weightsrs rG&test_sparse_design_with_sample_weightsr7^s ))   #Cs%8Lc7K )(A )AYYy)Q.2N"HX?OOAO7l00+2C2CQOrIc[R"SS/SS/SS/SS/SS//5n/SQn[SS9nURX5 g) Nrrrrrr)rWr[rr=)rArrr)rqrzrs rGtest_ridgecv_int_alphasr9tsK 4,q D$<#sc3ZPQAA < (E IIaOrIzparams, err_type, err_msgr>)rWrXiz alphas\[1\] == -1, must be > 0.0)grg$z"alphas\[0\] == -0.1, must be > 0.0)rWr1z1alphas\[2\] must be an instance of float, not strcSupE[RXE5n[RSSU5n[R"X#S9 U"S0UD6R Xg5 SSS5 g!,(df  g=f)z?Check the `alphas` validation in RidgeCV and RidgeClassifierCV.rrrKrNr)rprrrrr)rrOerr_typerrSrTrqrzs rGtest_ridgecv_alphas_validationr=}sZ(!I )(A Aq)$A x /F% 0 / /s A&& A4cSup[RX5nU[La[RU5nO[RSSU5nU"SS9R X45 g)zCheck the case when `alphas` is a scalar. This case was supported in the past when `alphas` where converted into array in `__init__`. We add this test to ensure backward compatibility. rrrKrWr=N)rprrrr)rrSrTrqrzs rGtest_ridgecv_alphas_scalarr?sU!I )(AG IIi  KK1i ( QA!rIc[SSS9nUR[[5 URR S[R S:Xdeg)Nr9rW)rrr)rrrrrr)rs rGtest_sparse_cg_max_iterrAsB {Q /CGGJ # 99??1 !1!1!!4 44 4rIz-ignore::sklearn.exceptions.ConvergenceWarningc Sn[[p![R"X S45Rn[ SS5HQnSHHn[ XTSS9nURX5 [UR[R"X@55 MJ MS SH.n[ USSS9nURX5 URcM.e g) NrKrWrR)r<r=r;r)rrr)r9r8r:r7) rrrAtilerhrangerrr.n_iter_)rrqrzy_nrrrs rG test_n_iterrGsI zq ''!^ $ & &C!QK-FveDC GGAO s{{BGGH,H I. 36A48 {{"""3rI)r;r9rrvwith_sample_weightcUS:Hn[SX$S9upVSnU(a>[RRU5nSUR UR SS9-nUS:XaS OUn [ U S US 9n [ US US 9n U RXVUS 9 U RU"U5XgS 9 [U RU R5 [U RU RS S9 g)aCheck that ridge finds the same coefs and intercept on dense and sparse input in the presence of sample weights. For now only sparse_cg and lbfgs can correctly fit an intercept with sparse X with default tol and max_iter. 'sag' is tested separately in test_ridge_fit_intercept_sparse_sag because it requires more iterations and should raise a warning if default max_iter is used. Other solvers raise an exception, as checked in test_ridge_fit_intercept_sparse_error rr)rTrVr-Nrrr`rvr9r)rrr-rgƠ>rW) r1rArcrdrfrrrr+rr) rrHrmrr-rqrzrrp dense_solverr5r4s rGtest_ridge_fit_intercept_sparserKs  H )$6 DAMii##$67ckkqwwqzk:: #)F"2;L|JKEHELOOA O6]1%qFK**L,C,CDK%%|'9'9ErI)r=r8r:c[SSS9up#U"U5n[US9nSRU5n[R"[ US9 UR XC5 SSS5 g!,(df  g=f)Nrr)rTrVrzsolver='{}' does not supportr)r1rformatrrrr)rrrqrzX_csrr4rs rG%test_ridge_fit_intercept_sparse_errorrOs^ *Ra HDA ! E'L,33F;G z 1" 2 1 1s A## A1c[SSUSS9up4U(a?[RRU5nSUR UR SS9-nOSnU"U5n[ SSS S S S 9n[S0UD6n [S0UD6n U RX4US 9 [R"5 [R"S[5 U RXtUS 9 SSS5 [U RU RSS9 [U RU RSS9 [ R""[SS9 [SS SSS9RXt5 SSS5 g!,(df  N=f!,(df  g=f)Nrrg@)rTrSrVr,rrr`r<Trr)r{rrrrrr-C6?rWz"sag" solver requires.*rrY)rrrrr)r1rArcrdrfrrrrrr simplefilterrr+rrrr) rHrmrrqrzrprrNrOr5r4s rG#test_ridge_fit_intercept_sparse_sagrSsG *1Cc DAii##$67ckkqwwqzk::  ! E %tF/&/K?6?LOOA O6  "g{3? #K**L,C,C$OK%%|'9'9E k)B C U$D4HLLUV D C # " D Cs',E2E( E%( E6return_interceptrr container)rvr9r:r;r<r=rc [S5nURSS5n/SQn[R"XV5nSnU(aSnXx- nU"U5n Sup[(aSOS n US :Hn US ;a:U(a3[ R "[S S 9 [U UU UUUU U S9 SSS5 g[U UU UUU UU S9nU(aUunn[XSU S9 [UUSU S9 g[XSU S9 g!,(df  g=f)z=check if all combinations of arguments give valid estimationsrrr)rWrKr7rg@)rYư>rYrQr)r<rvzIn Ridge, only 'sag' solverr)r{rrrTr-rN)r{rrr-rTrrr@r) r#rrArr2rrrrr+)rTrrUrrprq true_coefsrztrue_intercept X_testingr{rrr-outrrs rG.test_ridge_regression_check_arguments_validityr]s R C qAJ qAN A! IJE94$D H _$)9 ]]:-J K +!1! L   #)  Ciqt< >Ead;;L K s C'' C5)r8r9r:r;r<r=rcv[RRS5nSnUS:HnSupEURXE5nURU5nUR [R 5nUR [R 5n S[R "[R 5R-n [X SXS9n U RX5 U Rn [X SXS9n U RXg5 U RnU RUR:XdeURUR:XdeU RU5RUR:XdeU RU5RUR:Xde[U RU RSS S 9 g) NrrrrrK)r{rrrr-rQgMb@?rX)rArcrdrrAr6finfo resolutionrrrrCrr+)rrpr{r-rSrTX_64y_64X_32y_32rridge_32coef_32ridge_64coef_64s rGtest_dtype_matchrjSsh ))   "C E H I 99Y +D 99Y D ;;rzz "D ;;rzz "D bhhrzz"-- -CScH LLnnGScH LLnnG ==DJJ && & ==DJJ && &   D ! ' '4:: 55 5   D ! ' '4:: 55 5HNNHNNDIrIc2[RRS5n[R"SS/5nSup#nUR X#5nUR X$5nUR [R 5nUR [R 5n[USS9n U RXx5 U Rn [USS9n U RXV5 U Rn U RUR:XdeU RUR:XdeU RU5RUR:XdeU RU5RUR:Xde[U RU RSS9 g) Nrrr)rr rKr:rrr) rArcrdrrrAr6rrrrCrr,) rpr{rSrTn_targetrbrcrdrerfrgrhris rGtest_dtype_match_choleskyrmxsB ))   "C HHc3Z E&-#I8 99Y +D 99Y )D ;;rzz "D ;;rzz "D54H LLnnG54H LLnnG ==DJJ && & ==DJJ && &   D ! ' '4:: 55 5   D ! ' '4:: 55 5BrI)r8r:r;r9r<r=rseedc[RRU5nSup4URX45nURU5n[R"XV5SURU5--nSnUS:Hn [ 5n US:XaSOSn [R [R4H6n [URU 5URU 5UUUSU SS S S S 9 X'M8 U [R R[R :XdeU [RR[R:Xde[U [R U [RU S 9 g) Nrrrrr9rYrr_rF) r{rrVrr-rr return_n_iterrTr) rArcrdrrrr6r9rrArCr+) rrnrVrSrTrqrrzr{r-resultsr current_dtypes rG%test_ridge_regression_dtype_stabilityrssA 99((.L I91A   j )D q$!3!3I!>>>A E HfG[(4dD**bjj1 !1 HH] # HH] #%" " 2 2::  $ $ 22 2 2::  $ $ 22 2GBJJ')<4HrIc[SS9up[R"U5nUSSS2SS24nUSSS2n[SS9R X5 g)NrrVrKr<r)r rAasfortranarrayrr)rqrzs rGtest_ridge_sag_with_X_fortranrwsO  +DA !A #A#q& A #A#A A!rIzClassifier, paramscX[SSS9up#URSS5n[R"X3/SS9nU"S 0UD6R X$5nUR U5nUR UR :Xde[USS2S4USS2S45 [SS9R X#5 g) zNCheck that multilabel classification is supported and give meaningful results.rWr)rrVrXrNr<rr) r rrArrrrr.r) ClassifierrOrqrzrrrs rGtest_ridgeclassifier_multilabelrzs *AA FDA "aA vA&A  v  " "1 (C [[^F <<177 "" "vad|VAqD\2 A!rIrvr)rYrr7rcb[R"SS/SS/SS/SS//5n[R"SS /5nU(aS nURU5U-nOURU5n[US XS 9nUR X65 [R "UR S :5(deg)z:Test that positive Ridge finds true positive coefficients.rWrKrrRrrr rrZrTr{r-rrrN)rArrrrrer)rrr{rqrrrzrs rG#test_ridge_positive_regression_testr}s 1a&1a&1a&1a&12A 88QH D EE$K) # EE$K d6 E IIaO 66%++" # ## #rIc[RRS5nURSS5nUR SSUR SS9nU(a SnX4-U-nOX4-nXbR UR SS9S -- n/nS H8n[XUS S 9n URU RX65R5 M: [US SS.6 g)zTest that Ridge w/wo positive converges to the same solution. Ridge with positive=True and positive=False must give the same when the ground truth coefs are all positive. r,rr7rrWr`rr)TFr)r{r-rrrWr?N) rArcrdrrfrrgrrrrr+) rr{rprqrrrzrqr-rs rG%test_ridge_ground_truth_positive_testrs ))   #C #sA ;;sCaggaj; 1D Hy  H $t ++AG!-U  uyy,,- " W4a0rI)r8r:r;r9r<r=c Sn[R"SS/SS//5n[R"SS/5nX#-n[USUSS 9n[R"[ S S 9 UR X$5 S S S 5 [R"[ S S 9 [X$USUSS9u nS S S 5 g !,(df  NA=f!,(df  g =f)z5Test input validation for positive argument in Ridge.r7rWrKrrRrXTFr|zdoes not support positiverNzonly 'lbfgs' solver can be used)r-rrT)rArrrrrrr)rr{rqrrzrrxs rGtest_ridge_positive_error_testr s E 1a&1a&!"A 88QG D A V5 QE z)D E ! F z)J K %$v 1 L K F E L Ks B-B>- B;> C c^^ ^ [SSSS9um m SmSnS U UU 4Sjjn[TS9RT T 5n[TS S 9RT T 5nU"U5nU"U5nXV::de[U5HnU"XGS 9nXh::aMe g) z?Check ridge loss consistency when positive argument is enabled.rrrSrTrVr7rNcf>URnUbQ[RRU5nURUR SX RR S9-nO URnS[R"TTU-- U- S-5-ST-[R"US-5--$)Nrr`rrK)rrArcrdrrfrr) rrV noise_scalerrprrqr{rzs rG ridge_loss,test_positive_ridge_loss..ridge_loss$s$$  #))'' 5C;;Q ++BSBS!TTD;;DRVVQT\I5!;<> DNN LLOMGGA G.CIIt$/ 2 KKDqqwwqzKBMM"#bcFdNFGGA G. 99>> DNN GGAcrc1fIq"v]3B-?G@CIIt$/ 2 ?6? % %BEEF7O,C % DDHHQ!6H7   wvh&STUDJJ403# IIK . Q/0q 9B . Q/0 1B#((*O$i1n%*%nn &6 Q78qO# Q  b ! ?6?  q?  CD ?6?  r2_  EDDJJ +9rIrc>[SSUSS9up4[R"URS4S9nU(aSUSSS2'S n[ US US S 9nUR X4US 9 [R "/URQ[U5P7S9n[5n [U5Heup[U RX455HBun up[XS9nUR X=XMX]5 URX>5XSU 4'MD Mg [URU5 g)aNCheck that the predictions stored in `cv_results_` are on the original scale. The GCV approach works on scaled data: centered by an offset and scaled by the square root of the sample weights. Thus, prior to computing scores, the predictions need to be scaled back to the original scale. These predictions are the ones stored in `cv_results_` in `RidgeCV`. In this test, we check that the internal predictions stored in `cv_results_` are equivalent to a naive LOO-CV grid search with a `Ridge` estimator. Non-regression test for: https://github.com/scikit-learn/scikit-learn/issues/13998 rr[r)rSrTrrVrrNrKr rST)r>rVrrerr.)r rArrrremptyrr enumeraterhrrr+ri)rHrrrqrzrr>rrqrU alpha_idxr{idx train_idxtest_idxrs rG!test_ridge_cv_results_predictionsrs" "  DAGG1771:-0M cc F(# H  LL]L3((!8177!8CK!89K B%f- *3BHHQN*C &C&)CE IIalAL-2J K/4}}Q[/IKS)+ ,+D. H((+6rIc [SUS9up[R"URS4S9n[ SSS9nUR XUS9 [ 5n[URS 9n[R"URU5VVs/sH-upxUR XX'X7S9RX5PM/ snn5n [UR[X)5*5 g s snnf) zCheck that `RidgeCV` works properly with multioutput and sample_weight when `scoring != None`. We check the error reported by the RidgeCV is close to a naive LOO-CV using a Ridge estimator. rKrrVrrrST)rVrerrN)r rArrrrr rrXsqueezerhrr+rr) rmrqrzrrrUrtraintest y_pred_loos rG'test_ridge_cv_multioutput_sample_weightrs Q5G HDAGG1771:-0M7$OH LL]L3 B  (E "xx{   +  IIah 8LI M U U  +  JH((+=a+L*LM s 4C' c ^ [SSS9upSm U 4Sjn[US9nURX5 [5n[ UR S9n[ R"URU5VVs/sH,upgURXX5RX5PM. snn5n[URT "X5*5 gs snnf) zECheck that `RidgeCV` works properly with a custom multioutput scorer.rKrrcX- S-n[R"USS9nURS:Xa[R"USS/S9*$U*$)NrKrrrW)r)rArBndimr)y_truerEerrors mean_errorss rG custom_error=test_ridge_cv_custom_multioutput_scorer..custom_error sN/a'ggf1-   q JJ{QF;; ;|rIc4>T"X RU55*$)zGMultioutput score that give twice more importance to the second target.)r)rrqrzrs rGcustom_multioutput_scorerJtest_ridge_cv_custom_multioutput_scorer..custom_multioutput_scorer sQ 1 1! 4555rI)rVrN) r rrr rrXrArrhrr+r) rqrzrrrUrrrrrs @rG'test_ridge_cv_custom_multioutput_scorerr s QQ 7DA689H LL B  (EKM88TU;W;KE18QX & . .qw 7;WJH((<+F*FG Xs23C metaestimator)enable_metadata_routingc.U"5R5 g)zTest that `RidgeCV` or `RidgeClassifierCV` with default `scoring` argument (`None`), don't enter into `RecursionError` when metadata is routed. N)get_metadata_routing)rs rG*test_metadata_routing_with_default_scoringr2 s O((*rIzmetaestimator, make_datasetczU"SSSS9up#URX#[R"URS5S9 g)zTest that `set_score_request` is set within `RidgeCV.fit()` and `RidgeClassifierCV.fit()` when using the default scoring and no UnsetMetadataPassedError is raised. Regression test for the fix in PR #29634.rrrrrrN)rrArr)rrrqrzs rG+test_set_score_request_with_default_scoringr; s8 #!" EDAa"''!''!**=>rI) rrrr[rWg*@rLrLTFFN)r itertoolsrnumpyrArscipyrsklearnrr sklearn.basersklearn.datasetsrr r r sklearn.exceptionsr sklearn.linear_modelr rrrrrsklearn.linear_model._ridgerrrrrrrsklearn.metricsrrrsklearn.model_selectionrrrr r!sklearn.preprocessingr" sklearn.utilsr#sklearn.utils._array_apir$r%r&r'r(r)-sklearn.utils._test_common.instance_generatorr*sklearn.utils._testingr+r,r-r.r/sklearn.utils.estimator_checksr0r1sklearn.utils.fixesr2r3r4r5r6r7SOLVERSrr load_diabetesdiabetesrrrrrrindrcrdrpr& load_irisrrrrHrLfixturermark parametrizerrrrrrrrrrrrrr rr!r1rEr'r]r`rur{rrrrrrrrrrrrrrsortedrrrrr rrrrr"r2r7r9r TypeErrorr=r?rArrGrKrOrSrrr]rjrmrDrsrwrzr}rrrrrrrrrrrrrIrGrs , 2HG/,S B 4#S  ! ! #! Jii   #$iiA C $3i#C*S/ JDKK%+'(C&)C&L7+4-8)#9,)#X7+4-8#?9,#?L7+4-8%29,%2P7+4-82+9,2+j7+4-8289,28j7+4-82+9,2+j7+4-8+dVn-DE3+./'/F9,/'d+ 4<>$0*@"23.9 D:4 D"NO*T5M:.9.:;P.$"NO*T5M:.93:;P3&     *Z*1 G F^ #+  (F+R  +R\eW%56 ~(EFWh$784-8$K99G7$KNK@eW%56 ~(EF2w/#;G0G7;G|+dVn-DE8?F?6r3$ % (/: E 24GH-- i!$5$79L#MNay)0*  0DN3 P,$T:7I$JKeAh/0+dVn-DE"F1L"eAh/0+dVn-DE 2F1 2   F)-/'  %'AB   %  R   R'G',',T*F3I,JKFLF."  .9D: D ?F2C DE0F08 A .0LM:T:7I$JKCLC8w0A&BC=D=(ay)w0A&BCD*(K0$!N/1a&1a&1AB^#n4~EVP C P"w0A&BC M "J0ST * +  1 } %  @   &!D"&w0A&BC"D""5 KL#M#$#IJ-e}=.9#F:>K#FL#>?.9#:@#-e}=.9W:>W8+eT];4*?@rxxj>&AB O4<CA< 4(7VN4HD74E*FG-+.H+! O$  12-?.?rI