Explainable machine learning models for Medicare fraud detection

Journal of Big Data

Table 12 Mean AUC values by classifier and number of features for ten iterations of fivefold cross validation, for classifying the Medicare Part B data

Features classifier	10	15	20	25	30	80
CatBoost	0.9346	0.9409	0.9493	0.9537	0.9539	0.9569
ET	0.7907	0.8080	0.8122	0.8255	0.8214	0.8409
LightGBM	0.8300	0.8176	0.8412	0.8679	0.8643	0.8477
Logistic Regression	0.8349	0.8207	0.8240	0.7874	0.7896	0.8166
Random Forest	0.8096	0.8245	0.8374	0.8525	0.8476	0.8643
XGBoost	0.9375	0.9399	0.9493	0.9521	0.9529	0.9561