Table 2.
6x3 nested cross-validation results.
| KNN | SVM | Log Reg | Light GBM | NB | RF | |
|---|---|---|---|---|---|---|
| Accuracy | 62.0%(5.2%) | 69.0%(4.6%) | 69.7%(3.8%) | 69.4%(5.6%) | 65.1%(5.4%) | 66.9%(2.8%) |
| Precision | 65.3%(7.3%) | 70.2%(3.2%) | 68.2%(4.9%) | 68.9%(6.9%) | 65.7%(6.3%) | 67.3%(6.1%) |
| Recall | 52.4%(9.7%) | 65.4%(9.8%) | 73.8%(3.5%) | 71.1%(11.8%) | 60.9%(14.8%) | 67.4%(9.8%) |
| F1 | 57.5%(7.4%) | 67.4%(6.4%) | 70.8%(3.9%) | 69.4%(6.9%) | 62.6%(9.8%) | 66.7%(4.3%) |
| ROC-AUC | 61.5%(5.0%) | 68.8%(4.7%) | 69.6%(3.9%) | 69.4%(5.8%) | 64.9%(5.8%) | 66.7%(3.2%) |
Note: KNN – K-Nearest Neighbors. SVM - Support Vector Machine. Log Reg - Logistic Regression. Light GBM - Light Gradient Boosting Machine. NB - Naïve Bayes. RF - Random Forest. Number in parenthesis is the the average standard deviation from the six outer folds of the nested cross validation.