LogCAD: An Efficient and Robust Model for Log-Based Conformal Anomaly Detection

Table 5

Experimental results on BGL_100K unstable log sequences.

Injection rate (%)	Classifier	Accuracy	Precision	Recall	F1	MCC
5	LR	0.463	1	0.12	0.214	0.25
	SVM	0.667	0.833	0.156	0.263	0.32
	NB	0.756	0.857	0.72	0.783	0.622
	LogRobust	0.488	0.833	0.2	0.323	0.328
	CP	0.756	0.857	0.72	0.783	0.622
	LogCAD	0.854	0.806	1	0.893	0.71
10	LR	0.631	0.571	0.125	0.205	0.296
	SVM	0.667	0.833	0.156	0.263	0.32
	NB	0.655	0.714	0.156	0.758	0.677
	LogRobust	0.488	0.833	0.2	0.323	0.328
	CP	0.683	0.8	0.64	0.711	0.556
	LogCAD	0.854	0.806	1	0.893	0.71
15	LR	0.631	0.571	0.125	0.205	0.296
	SVM	0.667	0.833	0.156	0.263	0.32
	NB	0.655	0.714	0.156	0.758	0.677
	LogRobust	0.463	1	0.12	0.214	0.225
	CP	0.756	0.857	0.72	0.783	0.622
	LogCAD	0.829	0.909	0.8	0.851	0.709
20	LR	0.631	0.571	0.125	0.205	0.296
	SVM	0.655	0.714	0.156	0.256	0.321
	NB	0.69	0.568	0.781	0.658	0.56
	LogRobust	0.414	1	0.04	0.077	0.126
	CP	0.683	0.75	0.72	0.735	0.55
	LogCAD	0.756	0.8	0.8	0.8	0.613