Unsupervised Outlier Model Selection on 39 tabular benchmark datasets benchmark leaderboard · SOTA2 Research

Unsupervised Outlier Model Selection on 39 tabular benchmark datasets

68.77APGreedy Oracle

Evaluation Results

Method	Links
Greedy Oracle2026.05		68.77	1	89.68	65.04	69.06
MetaEns2026.05		43.08	59.3	78.67	40.42	46.81
ELECT (Top-10)2026.05		41.17	83.2	77.85	38.56	45.46
ELECT (Top-1)2026.05		40.69	85.8	77.34	38.61	45.19
MetaOD2026.05		39.89	101	75.47	37.46	43.92
ELECT+Random2026.05		39.81	102.2	77.19	37.78	44.49
Mega Ensemble2026.05		39.7	100	77.37	37.82	44.43
IForest2026.05		38.58	117	76.99	36.19	43.37
Global Best2026.05		37.87	122.7	75.83	35.93	42.66
Random Ensemble2026.05		37.59	124.3	74.77	36.08	42.9
LOF2026.05		35.13	120.1	74.39	32.97	41.69
Singleton2026.05		34.95	147.9	70.81	33.04	40.75
LSCP2026.05		34.84	124.3	75.6	34.41	42.51
RandNet2026.05		34.6	170.6	68.65	31.89	39.27
DTE-C2026.05		31.44	199.7	74.69	30.83	39.71
ROBOD2026.05		31.35	208	66.65	29.73	36.21
LUNAR2026.05		30.24	172	65.66	30.26	37.45
DAGMM2026.05		29.58	221.6	66.76	30.13	36.81
TCCM2026.05		29.29	188.3	67.45	28.34	36.9
RDA2026.05		27.42	211.9	70.63	27.21	35.15
DeepSVDD2026.05		20.73	247.5	59.05	21.16	29.68