Research Article

A Progressive Combined Variable Selection Method for Near-Infrared Spectral Analysis Based on Three-Step Hybrid Strategy

Table 4

Prediction results of starch content in corn by different variable selection methods.

Variable selection methodsNumber of variablesnLVRc2RMSEC (%)Rp2RMSEP (%)

None70080.94119.890.88228.05
CARS2560.980 ± 0.00211.58 ± 0.520.971 ± 0.00413.69 ± 0.97
VCPA990.990 ± 0.0018.04 ± 2.510.987 ± 0.0029.28 ± 2.19
IRIV1660.964 ± 0.00415.65 ± 1.680.920 ± 0.00922.59 ± 1.97
SiPLS18280.991 ± 0.0027.60 ± 1.630.976 ± 0.01012.65 ± 1.89
IVSO102100.984 ± 0.00210.26 ± 1.340.965 ± 0.00614.92 ± 2.11
BOSS1390.991 ± 0.0017.74 ± 0.850.990 ± 0.0028.02 ± 1.09
IVSO-BOSS3180.987 ± 0.0018.92 ± 0.490.974 ± 0.00313.52 ± 0.67
SiPLS-IVSO9070.998 ± 0.0014.05 ± 0.680.995 ± 0.0015.55 ± 0.84
SiPLS-BOSS2760.999 ± 0.0012.35 ± 0.570.997 ± 0.0014.22 ± 0.52
CARS-IVSO-BOSS850.812 ± 0.02428.15 ± 3.540.0751 ± 0.05434.02 ± 3.87
SiPLS-VCPA-BOSS770.709 ± 0.0429.31 ± 2.140.711 ± 0.0329.81 ± 1.89
SiPLS-IVSO-IRIV2360.999 ± 0.0012.52 ± 0.430.999 ± 0.0023.01 ± 0.68
SiPLS-IVSO-BOSS880.999 ± 0.0012.40 ± 0.230.999 ± 0.0012.80 ± 0.35