-iK SrSSKrSSKrSSKJr SSKJr SSKJ r SSK J r J r J r SSKJr SSKJr SS KJr SS KJrJr SS KJrJr SS KJr SS KJr SSKJrJ r SSK!J"r"J#r# \"SSS9ur$r%\"\$\%SS9ur$r%\"5RM\$5r$/SQr'S1\RP"5VVs1sH upUSiM snn-r)SQSjr*\RVRYS\5S5r-Sr.\RVRYS/\#Q\"Q5S5r/Sr0\RVRYS\'5\RVRYS \5S!55r1S"r2\RVRYS#S$5S%5r3S&r4S'r5S(r6S)r7\RVRYS*S+S,/5S-5r8\RVRYS.\#5S/5r9\RVRYS0\'5S15r:S2r;S3r<\RVRYS4S S50\Rz"S6\R|/\R|S6//54S S50S6S7/S7S6//40S6S7/S8S9//4/5S:5r?\RVRYS.\#5S;5r@\RVRYS.\#5S<5rAS=rBS>rCS?rD\RVRYS@SASB/5\RVRYSCSSD/5SE55rESFrF\RVRYSGSHSI/5SJ5rG\RVRYSKSLSM/5SN5rH\RVRYSOS+S,/5SP5rIgs snnf)RzF Tests for HDBSCAN clustering algorithm Based on the DBSCAN test code N)stats)distance)HDBSCAN)CONDENSED_dtype_condense_tree _do_labelling)_OUTLIER_ENCODING) make_blobs)fowlkes_mallows_score)_VALID_METRICSeuclidean_distances)BallTreeKDTree)StandardScaler)shuffle)assert_allcloseassert_array_equal)CSC_CONTAINERSCSR_CONTAINERS ) n_samples random_state)r)kd_tree ball_treebruteautolabelcv[[U5[- 5nUS:Xde[U[5U:deg)N)lenset OUTLIER_SETr y)labels threshold n_clusterss U/var/www/html/venv/lib/python3.13/site-packages/sklearn/cluster/tests/test_hdbscan.pycheck_label_qualityr+)s6S[;./J ??  +i 77 7 outlier_typec[R[RS.UnSSS.Un[USn[USn[R 5nUS/US'X/US'[ 5RU5nURU:HR5un[USS/5 U"URU5R5un[USS/5 [[SS55[[S S 55-n [ 5RXY5n [U RURU 5 g ) G Tests if np.inf and np.nan data are each treated as special outliers. )infinitemissingc X:H$Nxr&s r*#test_outlier_data..9sr,c.[R"U5$r3)npisnanr5s r*r7r8:s  r,r probrrN)r:infnanr Xcopyrfitlabels_nonzerorprobabilities_listrange) r-outlier prob_checkr r< X_outliermodelmissing_labels_idxmissing_probs_idx clean_indices clean_models r*test_outlier_datarR/s4 FF66G (+J l +G 4E \ *6 2DIQ)Aq62&u';';TBKKM(1a&1q!%U1c](;;M)-- 89K{**EMM-,HIr,c[[5nUR5n[SSS9R U5n[ X5 [ U5 Sn[R"[US9 [SSS9R [5 SSS5 SnSUS 'S US '[R"[US9 [SS 9R U5 SSS5 g!,(df  NU=f!,(df  g=f) zm Tests that HDBSCAN works with precomputed distance matrices, and throws the appropriate errors when needed. precomputedT)metricrCz*The precomputed distance matrix.*has shapematchNz'The precomputed distance matrix.*valuesr)rr=r=)r=rrU) r rBrCr fit_predictrr+pytestraises ValueError)D D_originalr'msgs r*test_hdbscan_distance_matrixr`Os AAJ M 5 A A! DFA" 7C z -}40<)rrN)r rBrCr:r@rArrDr flatnonzerorrHr$rI) r missing_labelinfinite_labelrLrMr'rNinfinite_labels_idx clean_idxrQ clean_labelss r*#test_dbscan_clustering_outlier_datars &i09M&z27;NIFFA;IaLrvv;IaLFFBFF#IaL IMM) $E  $ $, $ ?F(?@)Aq62..)AB*QC0Ss_s+=+S'TTUI)-- 45K00l0KL|I%67r,c[SS[R"[RS50S9R [5n[ U5 g)z, Tests that HDBSCAN using `BallTree` works. rwrrr=)rUrzN)rr:r}rBr|rYr+rms r*!test_hdbscan_best_balltree_metricrs?C1D+Ek!n r,c[[[5S- S9R[5n[ U5R [ 5(deg)zw Tests that HDBSCAN correctly does not generate a valid cluster when the `min_cluster_size` is too large for the data. r=min_cluster_sizeN)rr#rBrYr$issubsetr%rms r*test_hdbscan_no_clustersrs< c!fqj 1 = =a @F v;   , ,, ,r,c<[S[[5S5Hyn[US9R [5nUVs/sH o"S:wdM UPM nn[U5S:wdMI[ R "[ R"U55U:aMye gs snf)zV Test that the smallest non-noise cluster has at least `min_cluster_size` many points rtr=rrrN)rIr#rBrrYr:minbincount)rr'r true_labelss r*test_hdbscan_min_cluster_sizers{ "!SVQ/*:;GGJ*0@&RKu& @ { q 66"++k237GG GG 0@s B Bcr[Rn[US9R[5n[ U5 g)z9 Tests that HDBSCAN works when passed a callable metric. rXN)r euclideanrrYrBr+)rUr's r*test_hdbscan_callable_metricrs,  F F # / / 2Fr,treerrc[SUS9nSn[R"[US9 UR [ 5 SSS5 g!,(df  g=f)zu Tests that HDBSCAN correctly raises an error when passing precomputed data while requesting a tree-based algorithm. rTrUrqz%precomputed is not a valid metric forrVN)rrZr[r\rDrB)rrr_s r*"test_hdbscan_precomputed_non_bruters: $ 7C 1C z -   . - -s A A csr_containerc,[5R[5Rn[ U5 U"[5nUR 5n[5RU5Rn[ X5 [RS4[RS44HupV[R 5nXWS'[5RU5Rn[ U5 US[US:XdeUR 5nXSS'[5RU5Rn[ X5 M Sn[R"[US9 [SS S 9RU5 S S S 5 g !,(df  g =f) z Tests that HDBSCAN works correctly when passing sparse feature data. Evaluates correctness by comparing against the same data passed as a dense array. r0r1rrrr z4Sparse data matrices only support algorithm `brute`.rVrrrN)rrDrBrEr+rCrr:r@rAr rZr[r\) r dense_labels _X_sparseX_sparse sparse_labels outlier_valr-X_denser_s r*test_hdbscan_sparser s29==#++L %a I~~HIMM(+33M|3(*vvz&:RVVY>#$ h/77 <7&Q AC z -{k:>>xH . - -s "F FrqcSS/n[SSUSS9up#[SS9RU5n[XRUR 5HupVn[ XVS S S 9 [ XWS S S 9 M [US[RSS 9R[5nURRSS:XdeUR RSS:Xdeg )z^ Tests that HDBSCAN centers are calculated and stored properly, and are accurate to the data. )rdrd)@rirr)rrcenters cluster_stdboth) store_centersr=g?)rtolatol)rqrrN) r rrDzip centroids_medoids_rrBr|)rqrH_rcentercentroidmedoids r*test_hdbscan_centersr-s :&G 1gSV WDA  ' + +A .C$'$N &qt<QT:%O 6AGGAJ  c!f >>   "a '' ' <<  a A %% %r,c[RRS5nURSS5n[ SSSSS9R U5n[R "USS 9up4[U5S:XdeXCS :HS :de[ SS SSS S9R U5n[R "USS 9up4[U5S:XdeXCS :HS:Xdeg)zK Tests that HDBSCAN single-cluster selection with epsilon works correctly. rrtr>rdeomT)rcluster_selection_epsiloncluster_selection_methodallow_single_cluster) return_countsrg ףp= ?r)rrrrrqN)r:random RandomStaterandrrYuniquer#)rng no_structurer' unique_labelscountss r*.test_hdbscan_allow_single_cluster_with_epsilonrCs ))   "C88C#L "%!&!  k,  IIfDAM }  "" " 2% & ++ +"&!&!  k,  IIfDAM }  "" " 2% &! ++ +r,cSS/SS/SS/SS//n[SU/SQSS9up[5RU5Rn[ [ U55[ S U;5- nUS :Xde[X25S : g ) z Validate that HDBSCAN can properly cluster this difficult synthetic dataset. Note that DBSCAN fails on this (see HDBSCAN plotting example) g333333g333333?r"i)皙?gffffff?皙?rr)rrrrrGz?N)r rrDrEr#r$intr )rrBr&r'r)s r*test_hdbscan_better_than_dbscanrds u~t}q!fq"g>G +  DA Y]]1  % %FS[!Cf $55J ??&$t+r,z kwargs, XrTr=rtr"rc<[SSS0UD6RU5 g)zc Tests that HDBSCAN works correctly for array-likes and precomputed inputs with non-finite points. min_samplesr=Nr4)rrD)rBkwargss r*test_hdbscan_usable_inputsrxs $$V$((+r,cU"[R"S55nSn[R"[US9 [ SS9R U5 SSS5 g!,(df  g=f)zX Tests that HDBSCAN raises the correct error when there are too few non-zero distances. )rrz#There exists points with fewer thanrVrTrXN)r:zerosrZr[r\rrDrrBr_s r*-test_hdbscan_sparse_distances_too_few_nonzerorsI bhhx()A /C z -}%))!, . - -s A A(c"[R"S5nSUSS2SS24'SUSS2SS24'XR-nU"U5nSn[R"[ US9 [ SS 9RU5 SSS5 g!,(df  g=f) zi Tests that HDBSCAN raises the correct error when the distance matrix has multiple connected components. )rr=Nr>z3HDBSCAN cannot be performed on a disconnected graphrVrTrX)r:rTrZr[r\rrDrs r*0test_hdbscan_sparse_distances_disconnected_graphrs AAbqb"1"fIAab"#gJ CCAaA ?C z -}%))!, . - -s B BcSnSn[R"[US9 [SUS9R [ 5 SSS5 [R"[US9 [SUS9R [ 5 SSS5 [ [[R5[[R5- 5n[U5S:aC[R"[US9 [SUSS9R [ 5 SSS5 gg!,(df  N=f!,(df  N=f!,(df  g=f) zJ Tests that HDBSCAN correctly raises an error for invalid metric choices. cU$r3r4)r6s r*r72test_hdbscan_tree_invalid_metric..sr,zV.* is not a valid metric for a .*-based algorithm\. Please select a different metric\.rVr)rqrUNrr) rZr[r\rrDrBrHr$rrrr#)metric_callabler_metrics_not_kds r* test_hdbscan_tree_invalid_metricrs"O  z -)O<@@C . z -+o>BB1E . #h445FQ ]]:S 1 iq0A B F Fq I2 1 . - - -2 1s#DD$(!D5 D!$ D25 Ec[[[5S-S9nSn[R"[ US9 UR [5 SSS5 g!,(df  g=f)zl Tests that HDBSCAN correctly raises an error when setting `min_samples` larger than the number of samples. r=)rz min_samples (.*) must be at mostrVN)rr#rBrZr[r\rD)rr_s r*!test_hdbscan_too_many_min_samplesrs@ c!fqj )C -C z -   . - -s A A"c[R5n[RUS'Sn[ SS9n[ R "[US9 URU5 SSS5 g!,(df  g=f)zi Tests that HDBSCAN correctly raises an error when providing precomputed distances with `np.nan` values. rz(np.nan values found in precomputed-denserTrXrVN) rBrCr:rArrZr[r\rD)X_nanr_rs r*"test_hdbscan_precomputed_dense_nanrsP FFHE&&E$K 4C  'C z -  . - -s A'' A5rTFepsilonrcBSn[UUSS/SS/SS//S9upE[5RU5n[URUR S9nUS-US-US-1nUS-SUS-S US-S0n [ UUU UUS 9n [[U55V s0sH!o[R"X[:H5SS_M# n n [[U55V s0sH oXU _M n n [R"U R5"U5n[X5 g s sn fs sn f) zJ Tests that the `_do_labelling` helper function correctly assigns labels. 0rr)rrrrtr"rr=condensed_treeclusterscluster_label_maprrN)r rrDr_single_linkage_tree_rrrHr$r:where vectorizer~r)global_random_seedrrrrBr&estrrrr'_yfirst_with_label y_to_labelsaligned_targets r*test_labelling_distinctrs= I 'F G G DA )-- C# !!C4H4HNA y1}i!mB3q6lKlvr233lKK\\+//215Nv.LKs (DDc@SnSn[R"SSUS4SSSUS4SS/[S 9n[UU1USUS-S0S SS 9nUS S:n[ U5[ US :H5:Xde[UU1USUS-S0S SS 9nUS U:n[ U5[ US :H5:Xdeg)z Tests that the `_do_labelling` helper function correctly thresholds the incoming lambda values given various `cluster_selection_epsilon` values. r>g?rtr=)r>r=rr=r)r>r"rr=)r>rrr=)dtypeTrvaluerN)r:arrayrrsum)r MAX_LAMBDArr' num_noises r*test_labelling_thresholdingr s IJXX :q !  :q !     N%$aQ:!"# Fw'!+I y>S2. .. . %$aQ:!"# Fw'*4I y>S2. .. .r,rrrc[RRS5nURS5n[U5nSn[R "[ US9 [SUS9RU5 SSS5 g!,(df  g=f)zCheck that we raise an error if the centers are requested together with a precomputed input matrix. Non-regression test for: https://github.com/scikit-learn/scikit-learn/issues/27893 r)drtz>Cannot store centers when using a precomputed distance matrix.rVrT)rUrN) r:rrr rZr[r\rrD)rrrBX_disterr_msgs r*0test_hdbscan_error_precomputed_and_store_centersr%sd ))   "C 8A  #FNG z 1}MBFFvN 2 1 1s A:: B valid_algorrc>[SUS9R[5 g)zTest that HDBSCAN works with the "cosine" metric when the algorithm is set to "brute" or "auto". Non-regression test for issue #28631 cosinerN)rrYrB)rs r**test_hdbscan_cosine_metric_valid_algorithmr5s 8z2>>qAr, invalid_algoc[SUS9n[R"[SS9 UR [ 5 SSS5 g!,(df  g=f)zxTest that HDBSCAN raises an informative error is raised when an unsupported algorithm is used with the "cosine" metric. rrzcosine is not a valid metricrVN)rrZr[r\rYrB)rhdbscans r*,test_hdbscan_cosine_metric_invalid_algorithmr?s: X>G z)G HA I H Hs A A)r)J__doc__numpyr:rZscipyr scipy.spatialrsklearn.clusterrsklearn.cluster._hdbscan._treerrr sklearn.cluster._hdbscan.hdbscanr sklearn.datasetsr sklearn.metricsr sklearn.metrics.pairwiser r sklearn.neighborsrrsklearn.preprocessingr sklearn.utilsrsklearn.utils._testingrrsklearn.utils.fixesrrrBr& fit_transform ALGORITHMSitemsr%r+mark parametrizerRr`rkrnrrrrrrrrrrrrrr@rrrrrrrr rrr)routs00r*r/s  "# ?'1H.0!F>Cb11q!!$1""1% d1B1H1H1JK1Jvqc'l1JKK 8 ):;J<J>50-/Q/Q./QR S "  ,>2$3-$N 078884 - H )[!9:;.9I:IDj1&2&*,B,( M "BHHq"&&kBFFA;-G$HI M "aVaV$45 q!fq!f ,,.9 -: -.9 -: - J0 /$?QH-!/.@!/H&/R:x*@A OB O'89B:B)[)ABCuLs*M