Table 5. Performance evaluation by classification model

Machine learning Evaluation Feature selection PCA PCA + SMOTE AE AE + SMOTE
Logistic regression Accuracy 0.73 0.70 0.67 0.62 0.50
Precision 0.77 0.73 0.68 0.68 0.51
Recall 0.87 0.89 0.68 0.84 0.54
F1 score 0.82 0.80 0.68 0.75 0.52
SVM Accuracy 0.71 0.71 0.61 0.61 0.50
Precision 0.75 0.73 0.60 0.68 0.51
Recall 0.86 0.93 0.70 0.82 0.68
F1 score 0.80 0.81 0.65 0.74 0.58
Random forest Accuracy 0.72 0.70 0.64 0.59 0.49
Precision 0.75 0.75 0.61 0.67 0.50
Recall 0.90 0.85 0.81 0.76 0.73
F1 score 0.82 0.79 0.70 0.72 0.60
AdaBoost Accuracy 0.70 0.70 0.68 0.61 0.49
Precision 0.70 0.72 0.65 0.32 0.48
Recall 0.98 0.94 0.72 0.19 0.43
F1 score 0.82 0.81 0.69 0.24 0.46
PCA, principal component analysis; SMOTE, synthetic minority oversampling technique; AE, autoencoder; SVM, support vector machine.