Hyperparameter Tuning of Machine Learning Algorithms Using Response Surface Methodology: A Case Study of ANN, SVM, and DBN

Table 10

Experiment 4 results for the four kernel functions.

Kernel function	Average MAE from the validation set
	Fold 1	Fold 2	Fold 3	Fold 4	Fold 5	Fold 6	Fold 7	Fold 8	Fold 9	Fold 10
ANOVA	1.1964	1.2969	1.1607	1.1763	1.2061	1.2283	1.3041	1.2912	1.2432	1.249
Dot	1.1021	1.1573	1.1056	1.1024	1.1803	1.0865	1.32	1.2016	1.2016	1.1701
Radial	1.2124	1.2124	1.1136	1.0758	1.1215	1.1565	1.258	1.1821	1.1479	1.1373
Epanechnikov	1.1963	1.1885	1.1272	1.0887	1.1485	1.1271	1.2944	1.2157	1.1494	1.2193

Bold values indicate the minimal average MAE for each data fold (Fold 1 to Fold 10).