Attack detection in water distribution systems using machine learning

Table 4 A comparison of the algorithms BATADAL dataset with multi-stage implementation

Rank	Name	No. attacks.	S	\(S_{TTD}\)	\(S_{CLF}\)	\(F_1\)	TPR	TNR	PPV	TP	FP	TN	FN
1	B1	7	0.9701	0.9650	0.9752	0.9700	0.9533	0.9970	0.9873	388	5	1677	19
2	QDA	7	0.9503	0.9584	0.9422	0.8945	0.9165	0.9679	0.8735	373	54	1628	34
3	B2	7	0.9491	0.9580	0.9402	0.8813	0.9214	0.9590	0.8446	375	69	1613	32
4	Ensemble	7	0.9464	0.9400	0.9529	0.9061	0.9361	0.9696	0.8779	381	53	1692	26
5	MD	7	0.9342	0.9297	0.9387	0.9006	0.9017	0.9756	0.8995	367	41	1641	40
6	B3	7	0.9267	0.9360	0.9174	0.9057	0.8378	0.9970	0.9855	341	5	1677	66
7	LOF	7	0.9258	0.9229	0.9286	0.8875	0.8821	0.9752	0.8930	359	43	1693	48
8	SOD	7	0.9157	0.9091	0.9223	0.8462	0.8993	0.9453	0.7991	366	92	1590	41
9	B4	6	0.8942	0.8570	0.9313	0.8894	0.8894	0.9732	0.8894	362	45	1637	45
10	B5	7	0.8015	0.8350	0.7679	0.5382	0.8575	0.6784	0.3921	349	541	1141	58
11	LDA	6	0.7745	0.7959	0.7532	0.6677	0.5111	0.9952	0.9630	208	8	1674	199
12	B6	7	0.7727	0.8850	0.6605	0.4829	0.3292	0.9917	0.9054	134	14	1668	273
13	OSVM	7	0.7721	0.7383	0.8060	0.7538	0.6167	0.9952	0.9691	251	8	1674	156
14	Naïve	7	0.7500	1.0000	0.5000	0.3261	1.0000	0.0000	0.1948	407	1682	0	0
15	B7	3	0.5344	0.4290	0.6398	0.4220	0.3956	0.8841	0.4522	161	195	1487	246