A data-driven prioritisation framework to mitigate maintenance impact on passengers during metro line operation

European Transport Research Review

An Open Access Journal

Table 4 Anomaly detection model’s parameters and performance

	Selected parameters				Performance measures
Asset ID	ν	γ	Threshold (%)	ρ	Anomaly detection precision (%)	True anomaly rate (%)	PPV	TNR
1	0.05	0.00322	10	0.053799	67	67	0.564956	0.964856
2	0.05	0.00458	15	0.054133	67	100	0.581549	0.968125
3	0.01	0.000001	40	0.011452	100	100	0.822188	0.996917
4	0.01	0.00112	20	0.011158	75	75	0.733333	0.995465
5	0.05	0.00000142	20	0.052545	100	100	0.669077	0.975273
6	0.02	0.00159	35	0.021406	100	60	0.815661	0.995486
7	0.05	0.00000579	15	0.049346	67	100	0.717913	0.983427
9	0.02	0.000001	35	0.022559	67	67	0.649317	0.989856
10	0.01	0.000001	20	0.010219	100	67	0.813356	0.997402