Table 4.

Performance Results of the Selected Models Developed with Rieth et al. Dataset

Fault	Optimal feature subset size	AUC	Accuracy	Detection rate	False negative rate	False alarm rate	Latency (min)
1	18	100.0	99.8	99.8	0.3	0.0	15
2	10	99.7	99.2	99.1	0.9	0.0	24
3	10	50.2	17.0	0.4	99.6	0.0	2442
4*	1	100.0	100.0	100.0	0.0	0.0	3
5	14	100.0	100.0	100.0	0.0	0.0	3
6*	2	100.0	100.0	100.0	0.0	0.0	3
7	4	100.0	100.0	100.0	0.0	0.0	3
8	12	99.0	94.5	93.4	6.6	0.0	54
9	2	54.3	55.1	55.9	44.1	49.4	15
10	15	87.9	80.1	77.6	22.4	7.5	72
11*	2	99.8	95.9	95.1	4.9	0.0	18
12*	5	99.9	99.4	99.3	0.7	0.0	15
13	8	95.5	87.3	84.8	15.2	0.0	90
14*	2	100.0	100.0	100.0	0.0	0.0	3
15	16	56.0	17.6	1.1	98.9	0.0	1635
16*	2	96.7	88.3	87.4	12.6	7.2	9
17	28	96.9	92.9	91.5	8.5	0.0	210
18	5	98.6	96.1	95.3	4.7	0.0	75
19	10	100.0	99.6	99.6	0.4	0.0	18
20	14	97.1	92.7	91.2	8.8	0.0	99

^*Imply that the end-model is selected for online implementation.