Prediction of functional properties of registered chickpea samples using FT-IR spectroscopy and chemometrics

Abstract

This study was performed to evaluate the classification and prediction capability of Fourier transform infrared (FT-IR) spectroscopy for functional properties of chickpea samples for the first time. Health related (total phenolic content, iron chelating activity, and free radical scavenging activity) and food processing related properties (water soluble protein content, water binding capacity and oil binding capacity) of water soluble extracts of twelve registered chickpea cultivars which were grown in different cultivation year and area (n = 36 × 2) were investigated. The cultivars were not classified well based on cultivar type, cultivation location and year. Partial least squares (PLS) regression models were constructed with cross validation (leave one out) technique to determine the relation between the spectral data and chemical analysis. Regression coefficients of calibration models were found higher than 97% and regression coefficients of validation models were calculated higher than 95% meaning that the models have successful prediction capacity. Root mean square error of calibration (RMSEC) models ranged from 0.03 to 51.06 and cross-validation (RMSECV) models ranged from 0.03 to 50.33. Prediction of total phenolic content of the chickpea samples was not very good as other functional properties with higher RMSEC and validation models.