Table 3.

Cotton yield estimation based on vegetation indices.

Feature selection	Data set	PLSR			Elastic-Net			KRR			SVR			MLP			ELM
		R2	RMSE (t.ha−1)	rRMSE (%)	R 2	RMSE (t.ha−1)	rRMSE (%)	R 2	RMSE (t.ha−1)	rRMSE (%)	R 2	RMSE (t.ha−1)	rRMSE (%)	R 2	RMSE (t.ha−1)	rRMSE (%)	R 2	RMSE (t.ha−1)	rRMSE (%)
P	Cal	0.35	2.1833	77.56	0.34	2.2054	78.35	0.47	1.9694	69.96	0.38	2.1379	75.45	0.37	2.1512	76.42	0.56	1.8191	63.70
	Val	0.26	2.6585	76.59	0.27	2.6521	76.41	0.32	2.5701	74.04	0.28	2.6446	76.19	0.26	2.6484	76.30	0.52	2.0215	65.55
MIC	Cal	0.36	2.1768	77.33	0.35	2.1960	78.01	0.50	1.9288	68.52	0.38	2.1374	75.93	0.3	2.1600	76.73	0.67	1.5799	55.89
	Val	0.26	2.6597	76.62	0.27	2.6538	76.46	0.32	2.5866	74.52	0.27	2.6573	76.56	0.23	2.7400	78.94	0.61	1.7314	57.55
RF	Cal	0.36	2.1675	77.00	0.36	2.1836	77.57	0.61	1.7156	60.95	0.43	2.0644	73.34	0.43	2.0475	72.74	0.56	1.8812	63.04
	Val	0.29	2.6074	75.12	0.30	2.6120	75.25	0.39	2.4805	71.46	0.36	2.5325	72.96	0.33	2.5268	72.80	0.41	2.1842	69.46
RFE	Cal	0.37	2.1531	76.49	0.36	2.1819	77.51	0.56	1.8106	64.32	0.47	1.9710	70.02	0.50	1.9305	68.58	0.73	1.4285	47.57
	Val	0.27	2.6515	76.39	0.27	2.6559	76.52	0.39	2.4342	70.13	0.32	2.5926	74.69	0.33	2.5404	78.19	0.72	1.4418	49.23

The values in bold represent models with the best results in linear and nonlinear methods.