Table 6.
Majority vote classification accuracies. Majority vote decisions were developed from all ten motions, from only the top motions (EF, EE, P, S, WF, UD, and HC), and from a weighted majority vote. The best classification results within each feature set are in bold.
| Feature Set | Motions | Classification Accuracy (%) | ||
|---|---|---|---|---|
| LDA | SVM | RF | ||
| FS1 | All | 67.9 | 69.8 | 71.0 |
| Top | 70.4 | 69.1 | 75.3 | |
| Weighted | 72.2 | 71.0 | 74.1 | |
| FS2 | All | 70.4 | 58.6 | 69.8 |
| Top | 69.1 | 60.5 | 67.9 | |
| Weighted | 73.5 | 64.8 | 75.3 | |
| FS3 | All | 71.6 | 74.7 | 71.6 |
| Top | 72.2 | 72.8 | 74.1 | |
| Weighted | 71.6 | 73.5 | 77.2 | |
| FS4 | All | 77.8 | 73.4 | 62.3 |
| Top | 79.6 | 77.8 | 64.2 | |
| Weighted | 82.1 | 74.1 | 71.0 | |
| FS5 | All | 64.8 | 73.5 | 66.7 |
| Top | 68.5 | 76.5 | 65.4 | |
| Weighted | 67.3 | 75.9 | 75.3 | |
| FS5 Optimized with RELIEFF | All | 74.1 | 74.8 | 63.0 |
| Top | 74.1 | 77.2 | 62.3 | |
| Weighted | 79.6 | 81.5 | 77.2 | |