Skip to main content
. 2018 Jun;227:18–29. doi: 10.1016/j.jfoodeng.2018.01.009

Table 2.

Performance of the PLS regression model for moisture content in single green coffee beans.

Parameter Pre-processing LV R2 Cal RMSEC Slope R2 Val RMSECV Slope RPD
Natural moisture content
Log (1/R) 8 0.842 0.291 0.815 0.819 0.312 0.828 2.53
Mean centered 8 0.844 0.289 0.844 0.825 0.307 0.834 2.57
First derivative 11 0.899 0.233 0.899 0.858 0.276 0.879 2.86
Second derivative 12 0.898 0.237 0.898 0.813 0.322 0.852 2.45
SNV
12
0.897
0.235
0.897
0.850
0.285
0.872
2.77
Laboratory-treated beans Log (1/R) 5 0.951 2.234 0.951 0.950 2.279 0.949 4.73
Mean centered 5 0.947 2.314 0.947 0.947 2.372 0.946 4.54
First derivative 5 0.966 1.916 0.966 0.963 1.999 0.965 5.39
Second derivative 4 0.957 2.209 0.957 0.955 2.266 0.955 4.75
SNV 3 0.962 2.041 0.963 0.912 2.074 0.962 5.19

Spectral range used: 980-2480 nm. LV: latent variable. SNV: standard normal variate spectral treatment. Error is indicated as %. RPD: ratio of performance deviation. Sample size (n): 320.