Table 2 Validation performance of utilized machine learning models.
From: Semi-supervised learning framework for oil and gas pipeline failure detection
Model* | Class | Accuracy | False positive | False negative | Overall accuracy |
|---|---|---|---|---|---|
KNN | 1 | 56.4 | 41.3 | 43.6 | 59.0 |
2 | 53.8 | 45.5 | 46.2 | ||
3 | 53.8 | 36.6 | 33.3 | ||
Naïve bayes | 1 | 74.3 | 25.7 | 33.3 | 72.2 |
2 | 58.1 | 41.9 | 30.8 | ||
3 | 88.7 | 11.3 | 19.2 | ||
SVM | 1 | 78.2 | 21.8 | 21.8 | 76.1 |
2 | 73.1 | 36.0 | 26.9 | ||
3 | 76.9 | 10.4 | 23.1 | ||
CART | 1 | 96.2 | 3.8 | 2.6 | 88.9 |
2 | 86.8 | 13.2 | 15.4 | ||
3 | 83.5 | 16.5 | 15.4 | ||
ANN | 1 | 88.0 | 12.0 | 12.0 | 88.6 |
2 | 82.5 | 17.5 | 19.5 | ||
3 | 94.0 | 6.0 | 4.1 | ||
Boosted trees | 1 | 97.4 | 3.8 | 2.6 | 91.0 |
2 | 91.0 | 12.3 | 9.0 | ||
3 | 84.6 | 10.8 | 15.4 |
