Table 2.
Accuracy measures of different machine learning algorithms in predicting ‘use’ and ‘non-use’ of remote monitoring systems
| TP Rate (tp) | FP Rate (fp) | Precision | Recall | F-Measure | AUC | |
|---|---|---|---|---|---|---|
| ‘Use’(Decision Tree) | 85.7% | 11.1% | 85.7% | 85.7% | 85.7% | 0.76 |
| ‘Non-Use’(Decision Tree) | 88.9% | 14.3% | 88.9% | 88.9% | 88.9% | 0.76 |
| Overall (Decision Tree) | 87.5% | 12.9% | 87.5% | 87.5% | 87.5% | 0.76 |
| ‘Use’(k-NN) | 100% | 11% | 87.5% | 100% | 93.3% | 0.92 |
| ‘Non-Use’(k-NN) | 88.9% | 0% | 100% | 88.9% | 94.1% | 0.92 |
| Overall (k-NN) | 93.8% | 0.05% | 94.5% | 93.8% | 93.8% | 0.92 |
| ‘Use’(MLP) | 100% | 22% | 77.8% | 100% | 87.5% | 0.84 |
| ‘Non-Use’(MLP) | 77.8% | 0% | 100% | 77.8% | 87.5% | 0.84 |
| Overall (MLP) | 87.5% | 0.10% | 90.3% | 87.5% | 87.5% | 0.84 |
TP = true positive; FP = false positive; AUC = area under curve