Table 4 Confusion matrices of five ML models applied to the test dataset
(a) NB | |||
Total number of samples = 50 | Predicted failure mode |  | |
Rupture (22) | Leak (28) | ||
Actual failure mode | Rupture (23) | nTP = 19 | nFN = 4 |
Leak (27) | nFP = 3 | nTN = 24 | |
(b) SVM | |||
Total number of samples = 50 | Predicted failure mode | ||
Rupture (21) | Leak (29) | ||
Actual failure mode | Rupture (23) | nTP = 21 | nFN = 2 |
Leak (27) | nFP = 0 | nTN = 27 | |
(c) DT | |||
Total number of samples = 50 | Predicted failure mode | ||
Rupture (23) | Leak (27) | ||
Actual failure mode | Rupture (23) | nTP = 22 | nFN = 1 |
Leak (27) | nFP = 1 | nTN = 26 | |
(d) RF | |||
Total number of samples = 50 | Predicted failure mode | ||
Rupture (23) | Leak (27) | ||
Actual failure mode | Rupture (23) | nTP = 22 | nFN = 1 |
Leak (27) | nFP = 1 | nTN = 26 | |
(e) GB | |||
Total number of samples = 50 | Predicted failure mode | ||
Rupture (23) | Leak (27) | ||
Actual failure mode | Rupture (23) | nTP = 22 | nFN = 1 |
Leak (27) | nFP = 1 | nTN = 26 | |