A Variable Selection Method of Near Infrared Spectroscopy Based on Automatic Weighting Variable Combination Population Analysis

The automatic weighting variable combination population analysis (AWVCPA) for selection of characteristic variable in spectrum modeling is proposed, in which the low contribution wavelength variable is decreased by binary matrix sampling (BMS) and exponentially decreasing function (EDF). Taking near infrared spectra of beer and corn for example, the AWVCPA precision of prediction model is superior to those of other methods in the same situation.