Machine learning-driven prediction of medical expenses in triple-vessel PCI patients using feature selection

Table 4 Performance results of ML models on testing dataset

Testing Dataset
Methods	Number of variables	SMAPE	MAPE	MAE	MASE	RMSE	RSE	MSE
Variables selected by its own algorithm
LR	58	0.01098	0.01101	0.13619	0.30359	0.18750	0.17684	0.03516
RF	42	0.01119	0.01117	0.14093	0.27348	0.19081	0.13313	0.03641
SVR	45	0.01154	0.01157	0.14480	0.32277	0.23219	0.27117	0.05391
GLMBoost	30	0.00986	0.00984	0.12091	0.36158	0.16888	0.23111	0.02852
Bayesglm	37	0.01468	0.01466	0.18547	0.35989	0.25358	0.23513	0.06430
eXGB	28	0.00818	0.00816	0.10084	0.30155	0.15813	0.20264	0.02501
Variables averaged through six ML algorithms
LR	55	0.01098	0.01100	0.13616	0.30353	0.18741	0.17667	0.03512
RF	40	0.01129	0.01131	0.13983	0.31170	0.18920	0.18006	0.03580
SVR	38	0.01076	0.01074	0.13406	0.40091	0.24601	0.49046	0.06052
GLMBoost	32	0.00856	0.00856	0.10647	0.23733	0.16979	0.14501	0.02883
Bayesglm	35	0.01118	0.01120	0.13850	0.30874	0.18866	0.17903	0.03559
eXGB	25	0.00751	0.00755	0.09379	0.20908	0.15552	0.12166	0.02419

Abbreviations: LR Linear regression, RF Random forests, SVR Support vector regression, Bayesglm Bayesian generalized linear models, eXGB eXtreme gradient boosting, MAE Mean absolute error, MSE Mean square error, RMSE Root mean square error, RSE Relative Squared Error, MAPE Mean absolute percentage error, SMAPE Symmetric Mean Absolute Percentage Error, MASE Mean absolute scaled error

ISSN: 1472-6963