TABLE 5.

Performance of four ML methods and GAMLSS when applied to the OULAD. The best metrics are shown in bold characters (i.e., the lowest RMSE and MAE and the highest R ²). Means (M) and standard deviations (SD) are estimated across 10‐fold cross‐validation.

Method	RMSE		R 2		MAE
	M	SD	M	SD	M	SD
GAMLSS	0.1828	0.0061	0.0685	0.0291	0.1377	0.0038
RF	0.1803	0.0061	0.1061	0.0223	0.1364	0.0033
C&RT	0.1828	0.0067	0.0655	0.0168	0.1379	0.0043
nlSVM+k	0.1852	0.0070	0.0953	0.0168	0.1300	0.0038
EGB	0.1859	0.0075	0.0731	0.0225	0.1395	0.0051

Abbreviations: C&RT, classification and regression tree; EGB, extreme gradient boosting; GAMLSS, generalized additive models for location, scale, and shape; MAE, mean absolute error; ML, machine learning; nlSVM+k, nonlinear support vector machines with radial basis function kernel; OULAD, Open University Learning Dataset; RF, Random Forests; RMSE, root mean square error.