Table 2.
Comparison of validation accuracies of the best models trained on combinations of 5 questions, averaged over 10 combinations, using default hyperparameters, with demographics.
| Model | AUC Score | Standard Deviation | 95% Confidence Interval | F1 Score | Standard Deviation | 95% Confidence Interval |
|---|---|---|---|---|---|---|
| Logistic Regression | 87.79% | 0.54% | 87.58% - 87.97% | 87.76% | 0.55% | 87.55% - 87.95% |
| Gaussian NB | 86.16% | 0.41% | 86.02% - 86.30% | 86.05% | 0.42% | 85.92% - 86.20% |
| SVM | 87.66% | 0.70% | 87.40% - 87.93% | 87.61% | 0.71% | 87.35% - 87.89% |
| MLP | 87.93% | 0.87% | 87.44% - 88.01% | 87.89% | 0.87% | 87.40% - 87.97% |
| Random Forest | 89.75% | 0.72% | 89.52% - 89.96% | 89.67% | 0.71% | 89.43% - 89.88% |
| XGBoost | 87.82% | 0.96% | 87.46% - 88.18% | 87.80% | 0.93% | 87.43% - 88.14% |
| Ensemble | 89.82% | 0.84% | 89.60% - 90.07% | 89.78% | 0.82% | 89.56% - 90.03% |