Table 5.
Classification accuracies for each feature set. The best classification result for each motion within each feature set is in bold.
| Feature Set | Motions | Classification Accuracy (%) | ||
|---|---|---|---|---|
| LDA | SVM | RF | ||
| FS4 (MFL, MYOP) |
EF | 78.4 | 70.4 | 63.6 |
| EE | 68.5 | 65.4 | 67.3 | |
| P | 71.6 | 70.4 | 63.6 | |
| S | 72.2 | 71.0 | 70.4 | |
| WF | 70.3 | 70.4 | 70.4 | |
| WE | 66.0 | 60.5 | 57.4 | |
| UD | 77.2 | 79.6 | 67.3 | |
| RD | 74.7 | 69.1 | 67.9 | |
| HC | 69.8 | 69.8 | 63.6 | |
| HO | 63.0 | 70.4 | 63.6 | |
| FS5 (LOG, DASDV, MYOP, MAVS, PSR, AR4, ApEn, MFL, MSD) |
EF | 61.7 | 73.5 | 71.0 |
| EE | 75.9 | 72.8 | 72.8 | |
| P | 59.9 | 66.7 | 67.9 | |
| S | 58.0 | 71.0 | 67.9 | |
| WF | 64.8 | 59.9 | 69.1 | |
| WE | 51.9 | 58.6 | 63.0 | |
| UD | 69.1 | 74.7 | 64.8 | |
| RD | 61.1 | 73.5 | 67.3 | |
| HC | 56.2 | 63.6 | 67.3 | |
| HO | 56.8 | 64.8 | 60.5 | |
| FS5 Optimized with RELIEFF (PSR, MFL, MSD) |
EF | 69.8 | 71.6 | 64.8 |
| EE | 72.5 | 72.2 | 75.9 | |
| P | 72.8 | 70.4 | 66.0 | |
| S | 71.0 | 72.8 | 70.4 | |
| WF | 56.8 | 69.1 | 68.5 | |
| WE | 61.7 | 66.7 | 64.2 | |
| UD | 72.2 | 76.5 | 69.1 | |
| RD | 67.3 | 64.8 | 68.5 | |
| HC | 65.4 | 71.0 | 61.7 | |
| HO | 63.6 | 66.7 | 71.6 | |