Random Forest versus Support Vector Machine Models’ Applicability for Predicting Beam Shear Strength

Table 3

The prediction results for the investigated nine input combinations over the training and testing phases for the proposed SVM model.

Input combinations	R^2	RMSE (kN)	MAE (kN)	MAPE	Nash	MD
Training phase
M1	0.47	227.60	132.75	0.66	0.47	0.57
M2	0.63	209.86	75.61	0.27	0.55	0.76
M3	0.67	200.84	72.76	0.27	0.59	0.77
M4	0.77	174.56	60.52	0.28	0.68	0.80
M5	0.73	178.07	53.14	0.20	0.68	0.84
M6	0.76	167.05	43.83	0.18	0.72	0.87
M7	0.75	169.92	44.85	0.17	0.71	0.87
M8	0.74	170.11	45.06	0.18	0.71	0.87
M9	0.74	171.37	45.11	0.18	0.70	0.87
Testing phase
M1	0.59	226.27	106.65	0.49	0.47	0.62
M2	0.64	197.63	78.41	0.36	0.59	0.74
M3	0.72	187.49	72.20	0.37	0.63	0.76
M4	0.75	177.63	68.11	0.37	0.67	0.77
M5	0.77	174.56	60.52	0.28	0.68	0.80
M6	0.77	167.65	50.01	0.25	0.71	0.83
M7	0.78	166.61	49.59	0.29	0.71	0.84
M8	0.79	163.06	50.03	0.30	0.72	0.84
M9	0.79	163.17	51.20	0.34	0.72	0.83