Table 7.
Comparison of performance metrics of different machine learning models after introducing the balanced parameter (giving more weight to class 0). (Results of class 1.)
| Accuracy | Precision | Recall | F1-Score | |
|---|---|---|---|---|
| Train set | ||||
| Random Forest | 0.76 | 0.76 | 0.81 | 0.81 |
| XGBoost | 0.73 | 0.76 | 0.82 | 0.79 |
| SVM | 0.74 | 0.82 | 0.78 | 0.80 |
| Neural Network | 0.73 | 0.73 | 0.92 | 0.81 |
| Test set | ||||
| Random Forest | 0.76 | 0.81 | 0.80 | 0.80 |
| XGBoost | 0.67 | 0.68 | 0.74 | 0.71 |
| SVM | 0.66 | 0.74 | 0.74 | 0.74 |
| Neural Network | 0.65 | 0.65 | 0.86 | 0.74 |
Class 1 results; target variable: total PA higher or lower than 600 MET·min/week; features included in the model involved age categories, sex, BMI, crowding index, education, marital status, job location during lockdown, auto-declared PA change during lockdown, type of PA during lockdown, employment status, and sitting time.