used to develop a model by taking the predictor as X and predicted as y. In this case X is a spectral data and y is the wet gluten concentration for different combinations. In case of first set of data. 94 blocks were made containing two spectra of each combination. In PLSR modeling. each block was kept out once to make calibration model and then last block was predicted using technique. This process is repeated until all the blocks were taken out once to develop a more robust model. Similarly, the dataset was divided into 70:30 for calibration and validation respectively to determine the robustness of model. The performance of the models was evaluated by using high value of correlation coefficient with low value of error.