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

Figure 3

Results of fine selection by iterative variable subset optimization for oil content prediction in corn: (a) the root mean square error of cross-validation varying with the number of sampling and (b) the number of variables selected varying with the number of sampling. The lowest root mean square error appeared in the fourth iteration with 52 variables selected.

(a)

(b)