Fig. 2. The cost of the designed monitoring program (a) and the percentage of cost reduction (b) using four different machine learning algorithms (including Extreme gradient boosting (XGB), Decision tree (DT), Logistic regression (LR), and Support vector machine (SVM)), each with default and tuned model parameters.

Default_LR, Default_SVM, Default_DT, and Default_XGB represent ML algorithms using default parameter settings. Tuning_LR, Tuning_SVM, Tuning_DT, and Tuning_XGB represent ML algorithms using tuned parameter settings. AUC_train, AUC_validation, and AUC_test, represent AUC score on train, validation, and test dataset. Cost_train, Cost_validation, and Cost_test, represent monitoring cost on train, validation, and test dataset. Cost_train_reduction, Cost_validation_reduction, and Cost_test_reduction, represent monitoring cost reduction on train, validation, and test dataset.