Prediction of Yellowness Index Using Partial Least Square Regression Model

Partial least square regression (PLSR) is a multivariate statistical method that has been widely used in various industries such as chemical, food, agriculture, and plastic industries. However, a limited study utilizing the PLSR model to predict the colour indexes including yellowness index (YI). YI is a value determined from spectrophotometric data which assesses perceptual yellowness. In industries like paints, and plastics, yellowness is considered as it relates to the quality of the products that is degraded by light, chemical exposure, and processing. The spectrophotometric data for YI calculation is normally measured by high-priced and space-occupied hardware instruments. Moreover, a direct YI measurement using these instruments are not possible. Hence, this study aims to investigate the predictive performance of the PLSR model on YI measurement and its results are compared with the principal component regression (PCR). The data is generated using a standard colour chart with the RGB value listed. It was found that the root mean square errors (RMSE) of PLSR for training and testing datasets are lowered by 12.55% and 5.68%, respectively as compared to PCR. Furthermore, its coefficients of determination (R²) are also higher than PCR and the R² values of PLSR are high, which are more than 0.9.