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

Figure 5

Variable selection results of oil spectral data in corn based on method combined synergy interval partial least squares, iterative variable subset optimization, and bootstrapping soft shrinkage. The curve was average spectral data of 80 corn samples with 700 variables from 1,100 to 2,498 nanometers. Variables selected in the first, second, and third steps were marked in the figure with different symbols. The distribution trend of variables could be seen by comparing the average spectrum.