Table 4.

Cotton yield estimation results based on texture features.

Feature selection	Data set	PLSR			Elastic-Net			KRR			SVR			MLP			ELM
		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 (%)	R 2	RMSE (t.ha−1)	rRMSE (%)
P	Cal	0.29	2.2891	81.32	0.28	2.3116	82.12	0.43	2.0515	72.88	0.33	2.2290	79.19	0.30	2.2644	80.44	0.72	1.4666	50.09
	Val	0.27	2.6575	76.56	0.24	2.7222	78.43	0.41	2.3907	68.88	0.33	2.5235	72.70	0.24	2.6975	77.72	0.67	1.5518	57.96
MIC	Cal	0.28	2.2989	81.67	0.28	2.3107	82.09	0.31	2.2599	80.28	0.30	2.2868	81.24	0.29	2.2897	81.34	0.66	1.7020	53.09
	Val	0.25	2.6807	77.23	0.25	2.7053	77.94	0.36	2.5190	72.57	0.33	2.5240	72.72	0.29	2.6043	75.03	0.57	1.8765	56.86
RF	Cal	0.52	1.8897	67.13	0.50	1.9702	67.00	0.60	1.7164	60.97	0.57	1.7760	63.09	0.57	1.7789	63.19	0.77	1.3601	42.93
	Val	0.44	2.3066	66.45	0.43	2.3906	68.87	0.59	2.0112	57.94	0.58	2.0398	58.77	0.54	2.0940	60.33	0.74	1.4065	44.42
RFE	Cal	0.22	2.4007	85.28	0.20	2.4267	86.21	0.38	2.1502	76.38	0.24	2.3826	84.68	0.24	2.3685	84.14	0.86	1.0941	38.46
	Val	0.16	2.8485	82.07	0.12	2.9104	83.85	0.34	2.5443	73.30	0.18	2.8473	82.03	0.17	2.8186	81.20	0.83	1.1705	34.54

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