:i*^SSKJr SSKrSSKrSSKJrJr SSKr SSK r SSK r SSKr SSKrSSKJr SSKJr SSKJr \(aSS KJr SS jrS r\S 5r\SS j5r\S5r\S5r\S5rSSjrSSSjjr SSSjjr!g)) annotationsN) TYPE_CHECKINGLiteral)njit)DimensionError) show_progress) _ArrayLikecLU[RR"UR6S--n[R R RURUR55RUS9n[RRRU5$)N:0yE>metric)nprandomrandnshapescipyspatialdistancepdistfillnameanTcluster hierarchycomplete)Xr X_full_rankDs Y/srv/projetos/modelo_ml_acdoc/venv/lib/python3.13/site-packages/shap/utils/_clustering.pypartition_treer"swbiiooqww/$66K $$[%7%7 8H8H8J%K%M%MV\$]A == " " + +A ..cV[U5n[URSS- XX#S5 g)aRandomly shuffle the indexes in a way that is consistent with the given partition tree. Parameters ---------- indexes: np.array The output location of the indexes we want shuffled. Note that len(indexes) should equal index_mask.sum(). index_mask: np.array A bool mask of which indexes we want to include in the shuffled list. partition_tree: np.array The partition tree we should follow. rr N)len_pt_shuffle_recr)indexes index_maskr"Ms r!partition_tree_shuffler*s, JAN((+a/nYZ[r#c<US:aX U-(a X-X'US-$U$[X0S4U- 5n[X0S4U- 5n[RR5S:a[ XaX#XE5n[ XqX#XE5nU$[ XqX#XE5n[ XaX#XE5nU$)Nrr )intrrrr&)ir'r(r"r)posleftrights r!r&r&-s1u !e 5GL7NJ ~d#a' (D !t$q( )E yy1dZPej!Q Jej!QdZP Jr#c [R"[U55n[U5H\n[ [SU55H@n[S[U5U- 5H!n[ XXe5S:dM[ X6U5 M# MB M^ U$)Nrr r)raranger%rangelist_reverse_window_score_gain_reverse_window) all_masks max_swap_size num_passesorder_lengthr-s r!delta_minimization_orderr=Ass IIc)n %E : 5M23F1c%j612-iJQN#Ef534 Lr#cx[US-5H(nXU-nXU-U- S- XU-'X@X-U- S- 'M* g)Nrr )r3)r:startr<r-tmps r!r6r6LsR 6Q; AI !!3a!78ai(+enq 1$% r#c[XUS- XU5[XX#-S- XX#-5-n[XUS- XX#-S- 5[XUXX#-5-nXE- $)Nr )_mask_delta_score)masksr:r?r< forward_score reverse_scores r!r5r5Ts%e%!),<&=u5\?RSVg ENQ&'(%en0E*FWM&e%!),<&=u5>\]K]E^?_`ct ElU#89dM  ((r#c&X- R5$)N)sum)m1m2s r!rBrB`s G==?r#cF[RRRX5n[RR R U5n[RR R[RR RXC55$)z\A leaf ordering is under-defined, this picks the ordering that keeps nearby samples similar.) rrrrrrr leaves_listoptimal_leaf_ordering)rr anchor_firstr cluster_matrixs r!hclust_orderingrOesi  $$Q/A]],,55a8N == " " . .u}}/F/F/\/\]k/o ppr#cSSKn[RRXUS9uppURSn /n /n[ U 5HnUR UUUSUS9nURUSS2XS-24XSS2XS-24U 4/SS9 U RURUSS2XS-2455 URURU SS2XS-2455 M [R"U 5Rn[R"U5Rn[R"X45n[[R "[ U 5[ U 55X-S9HunnUU:XaM[R""USS2U45nUS :a[$R&"S US 35 SnOUR UUUSUS9nURUSS2UUS-24USS2U4U SS2UUS-24USS2U44/SS9 [)SS[R*"USS2U4URU SS2UUS-245- S -5U- - 5nSU- UUU4'M U$) aCompute redundancy distances scaled from 0-1 among all the features in X relative to the label y. Distances are measured by training univariate XGBoost models of y for all the features, and then predicting the output of these models using univariate XGBoost models of other features. If one feature can effectively predict the output of another feature's univariate XGBoost model of y, then the second feature is redundant with the first with respect to y. A distance of 1 corresponds to no redundancy while a distance of 0 corresponds to perfect redundancy (measured using the proportion of variance explained). Note these distances are not symmetric. Returns ------- np.ndarray A square matrix of shape (n_features, n_features) containing the pairwise redundancy distances between features. Each element [i, j] represents the redundancy distance from feature i to feature j with respect to y. rN random_stater ) subsample n_estimators learning_rate max_depthearly_stopping_roundsF)eval_setverbose)totalg-C6?z!No/low signal found from feature z (this is typically caused by constant or near-constant features)! Cluster distances can't be computed for it (so setting all redundancy distances to 1).r)xgboostsklearnmodel_selectiontrain_test_splitrr3 XGBRegressorfitappendpredictrvstackrzerosr itproductvarwarningswarnmaxr)ryrUrWrSmax_estimatorsrRr[X_trainX_testy_trainy_test num_featurestrain_preds_listtest_preds_listr-model train_preds test_predsdistj preds_varr2s r!xgboost_distances_r2r{ms4(/'>'>'O'OPQco'O'p$GW771:LO < $$''"7 %   '!QQY,'Aqq5yL=QSYxx " = MM. !& yq)*A MM)AqD/rzz)AqD//JQV W+}}%%++IKK"))//9;;K\K\:]`d:d,dms+t(}}&&--d33 J }}&&//55 I }}&&..t44 r#) correlation)dr) sqeuclideanF)g333333?rg?i'r) rUfloatrW int | NonerSz float | NonerlrrRint | np.random.RandomStatereturn np.ndarray)Nr}rr) rr rkz_ArrayLike | Nonerz(Literal['single', 'complete', 'average']rstrrRrrr)" __future__r itertoolsrerhtypingrrnumpyrpandasr scipy.clusterr scipy.spatialr\numbarutils._exceptionsr_show_progressr _typesr r"r*r&r=r6r5rBrOr{rr#r!rsP").)"/ \$&,,))q()!!'01VV& V  V  V.VVv 8@01 Z5Z5Z56Z5  Z5 . Z5  Z5r#