Research Article

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

Table 5

Prediction results of organic matter content in soil by different variable selection methods.

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

None1,050100.976184.250.749318.79
CARS7280.985 ± 0.002146.77 ± 10.150.916 ± 0.022154.80 ± 22.36
VCPA870.978 ± 0.016151.41 ± 10.530.948 ± 0.031122.40 ± 20.24
IRIV1190.983 ± 0.006149.90 ± 12.130.933 ± 0.027139.06 ± 15.47
SiPLS14080.985 ± 0.007143.04 ± 15.890.876 ± 0.035240.06 ± 24.19
IVSO20280.982 ± 0.009152.96 ± 16.430.923 ± 0.029148.41 ± 24.18
BOSS50100.987 ± 0.003134.07 ± 24.730.901 ± 0.024168.21 ± 25.12
IVSO-BOSS2660.985 ± 0.003145.44 ± 21.310.891 ± 0.187193.45 ± 22.74
SiPLS-IVSO6880.987 ± 0.001137.22 ± 10.740.977 ± 0.16895.41 ± 12.54
SiPLS-BOSS1370.988 ± 0.001131.71 ± 11.850.971 ± 0.013108.80 ± 20.57
CARS-IVSO-BOSS680.901 ± 0.027200.57 ± 20.640.682 ± 0.058347.23 ± 24.13
SiPLS-VCPA-BOSS960.984 ± 0.017142.87 ± 13.540.941 ± 0.067204.19 ± 27.10
SiPLS-IVSO-IRIV1580.983 ± 0.002131.71 ± 12.640.977 ± 0.029100.04 ± 16.47
SiPLS-IVSO-BOSS1090.988 ± 0.001129.99 ± 10.070.979 ± 0.00994.03 ± 11.76