Prediction of Soil Moisture-Holding Capacity with Support Vector Machines in Dry Subhumid Tropics

Table 4

Calibration results of optimal hyperparameters of SVR model types.

	Predictors	C		ε		SVs		CV errors
		1st	2nd	1st	2nd	1st	2nd	1st	2nd
FC1	Sand + clay	100	127	0.197	0.21	151	143	0.043	0.044
FC2	Clay + sand + BD	1	0.1	0	0.25	198	128	0.034	0.034
FC3	Clay + sand + BD + OC	100	96	0.006	0.022	196	189	0.034	0.034
FC4	Sand + BD	1	0.4	0.244	0.24	126	128	0.034	0.034
WP1	Sand + clay	0.1	0.35	0.199	0.217	146	138	0.080	0.081
WP2	Clay + sand + BD	1000	950	0.246	0.24	119	124	0.067	0.067
WP3	Clay + sand + BD + OC	0.1	0.65	0.15	0.15	151	149	0.065	0.065
WP4	Sand + BD + OC	10	28	0.037	0.156	183	150	0.033	0.065