A New Method for Spectral Wavelength Selection Based on Multiple Linear Regression Combined with Ant Colony Optimization and Genetic Algorithm

Table 4

Results of PLS modeling with different methods on different datasets.

Dataset	Method	RMSECV^a	RMSECV^b
Wheat	Full	0.4305	—
	CARS	0.4188  0.0142	0.3893–0.4356
	MLR-ACO	0.3654  0.0034	0.3607–0.3732
	MLR-ACO-GA	0.3850  0.0243	0.3655–0.4517
Grain	Full	0.4433	—
	CARS	0.3321  0.0298	0.2889–0.4293
	MLR-ACO	0.2779  0.0119	0.2576–0.3151
	MLR-ACO-GA	0.2874  0.0186	0.2652–0.2889
Corn	Full	0.1108	—
	CARS	0.0230  0.0042	0.0168–0.0372
	MLR-ACO	0.0245  0.0078	0.0115–0.0424
	MLR-ACO-GA	0.0239  0.0070	0.0099–0.0392

^aStatistical results of mean ± standard deviation of 30 replicate simulations of different methods. ^bStatistical results of the variation range of 30 repeated simulations with different methods.