Table 4.
Averaged automatic scoring performance for the eight models on the training data
| Model | Summed scores | Rounded average scores | ||||
|---|---|---|---|---|---|---|
| ± 2 Accuracy | ± 4 Accuracy | Correlation | ± 0 Accuracy | ± 1 Accuracy | Correlation | |
| Random | 0.28 | 0.47 | < 0.01 | 0.27 | 0.69 | 0.01 |
| Median | 0.35 | 0.62 | NaN | 0.33 | 0.86 | NaN |
| T | 0.97 | 1.00 | 0.91 | 0.88 | 1.00 | 0.86 |
| TG | 0.95 | 0.99 | 0.95 | 0.86 | 1.00 | 0.92 |
| TM | 0.98 | 1.00 | 0.96 | 0.88 | 1.00 | 0.93 |
| TMV | 0.94 | 0.99 | 0.98 | 0.85 | 1.00 | 0.94 |
| TMG | 0.89 | 0.99 | 0.97 | 0.84 | 1.00 | 0.93 |
| TMVG | 0.77 | 0.98 | 0.98 | 0.73 | 1.00 | 0.94 |
The abbreviations indicate which features are included in each model: T time, G grip angle, M tip motion, and V tool visibility. NaN signifies “not a number” and occurs because correlation with a constant rating is undefined