Table 2. Model accuracy of the included machine learning algorithms Based on confusion matrix parameters.
| Confusion matrix Parameters (%) | The included machine-learning algorithms | |||||||
|---|---|---|---|---|---|---|---|---|
| PART | Naïve Bayes | Random forest | Logit Boost | J48 | AdaBoost | Multilayer perceptron | LR | |
| True positive rate (%) | 89.90 | 64.00 | 88.50 | 72.10 | 88.40 | 69.80 | 83.60 | 70.90 |
| False positive rate (%) | 18.20 | 30.80 | 1.50 | 38.90 | 32.40 | 35.21 | 12.10 | 36.30 |
| Precision (%) | 93.80 | 73.90 | 87.80 | 70.60 | 76.00 | 72.00 | 81.00 | 71.70 |
| F-measure (%) | 94.30 | 68.20 | 88.60 | 71.30 | 77.71 | 70.90 | 82.30 | 71.30 |
| Relative absolute error (%) | 51.78 | 75.33 | 34.05 | 83.68 | 75.05 | 85.10 | 23.19 | 84.21 |
| AUC (%) | 91.89 | 72.30 | 82.70 | 73.20 | 86.01 | 72.90 | 83.29 | 65.60 |
| Kappa statistics (%) | 86.57 | 32.66 | 78.68 | 33.26 | 79.27 | 36.50 | 72.02 | 35.00 |
| Accuracy (%) | 95.53 | 66.29 | 82.37 | 77.28 | 89.24 | 68.60 | 87.20 | 65.80 |
| Note that LR, stands for Logistic Regression | ||||||||