Table A1.
The accuracy of the test datasets according to the selected machine learning classifiers. “PN” refers to the positive/negative movement model, “FP” denotes the fast positive/undefined movement model, and “FN” represents the fast negative/undefined movement model.
| Classifier’s Name | Region and Dataset’s Name | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Lombardy Dataset | Lisbon Dataset | Washington Dataset | |||||||
| PN | FP | FN | PN | FP | FN | PN | FP | FN | |
| Cosine K-NN | 75.5% | 71.8% | 85.8% | 80.6% | 82% | 81.4% | 87.4% | 78% | 80.7% |
| Subspace K-NN | 72.7% | 72.3% | 84.7% | 77.1% | 82.3% | 80.3% | 84.4% | 79.8% | 80.7% |
| Medium Neural Network | 72.4% | 70.2% | 82.8% | 76% | 79.6% | 78.1% | 82.9% | 74.8% | 80.3% |
| Logistic Regression | 67.5% | 72.3% | 82.8% | 75.3% | 80.7% | 79.5% | 85.9% | 74% | 79.4% |
| Cubic SVM | 74.8% | 70.2% | 84.9% | 79.9% | 82% | 81.3% | 86.6% | 79% | 80.8% |
| Medium Tree | 63.8% | 69.7% | 80% | 71.9% | 78.3% | 75.2% | 82.4% | 74.8% | 77.5% |
| Fine Tree | 67.3% | 62.6% | 82.3% | 75.1% | 79.6% | 77.6% | 83.6% | 75.2% | 80.8% |
| Bagged Tree | 74.1% | 71.8% | 85.3% | 79.3% | 79.9% | 81.1% | 85.6% | 79% | 81.3% |
| Quadratic Discriminant | 71.7% | 75.2% | 81.6% | 79.8% | 81.1% | 80.5% | 88.5% | 81.6% | 79.6% |
| 2D Convolution Neural Network | 67.1% | 68.2% | 82.4% | 77.8% | 83.5% | 80.5% | 80.4% | 68.2% | 79.8% |