One-Class Classification by Ensembles of Random Planes (OCCERPs)

Table 4

Average AUCROC of various OCC algorithms against the OCCERP (500) algorithm on various domain datasets presented in Table 2. Bold numbers indicate the best performance.

Dataset	If	LOF	OCSVM	Autoencoder	OCCERP (500)
MF	0.969	0.890	0.978	0.941	0.992
COV	0.831	0.912	0.804	0.769	0.883
DLR	0.947	0.988	0.955	0.978	0.991
Class-level-kc1-defectornot	0.797	0.762	0.705	0.607	0.801
kc2	0.839	0.632	0.806	0.754	0.827
kc1	0.792	0.634	0.708	0.634	0.807
cm1	0.704	0.661	0.636	0.518	0.787
Datatrieve	0.728	0.690	0.692	0.572	0.753
pc1	0.697	0.689	0.676	0.599	0.719
Class-level-kc1	0.903	0.884	0.864	0.780	0.891
-Defect-count-ranking
Best performance	2	1	0	0	7