Research Article
Supporting Risk-Aware Use of Online Translation Tools in Delivering Mental Healthcare Services among Spanish-Speaking Populations
Table 3
Performance of RVM classifiers with lexical sentiment features.
| RVM | Training data | Testing data | AUC mean | SD | AUC | Accuracy | Sensitivity | Specificity |
| LIWC all 92 | 0.614 | 0.057 | 0.580 | 0.548 | 0.518 | 0.577 | LIWC all 92 with min-max | 0.577 | 0.054 | 0.646 | 0.619 | 0.566 | 0.671 | LIWC all 92 with L2 | 0.610 | 0.064 | 0.573 | 0.548 | 0.518 | 0.577 | LIWC all 92 with Z-score | 0.619 | 0.046 | 0.670 | 0.619 | 0.506 | 0.729 | L10 (automatically optimised) | 0.602 | 0.040 | 0.605 | 0.571 | 0.651 | 0.494 | L10 with min-max | 0.629 | 0.055 | 0.607 | 0.566 | 0.566 | 0.565 | L10 with L2 | 0.604 | 0.034 | 0.609 | 0.571 | 0.518 | 0.624 | L10 with Z-score | 0.614 | 0.068 | 0.616 | 0.583 | 0.578 | 0.588 |
|
|