Non-Destructive Sugar Assessment in Cashew Apples Using Reflectance Spectra

Sugar content in cashew apples is a critical indicator of fruit quality and maturity, directly influencing processing and market value. This study explores the use of spectral indices derived from reflectance data for the non-destructive prediction of sugar content (˚Brix) in cashew apples. Two multivariate modeling approaches were evaluated: Partial Least Squares Regression (PLSR) and Random Forest (RF). The PLSR model, optimized with 8 components, achieved high predictive performance, with R2 = 0.9313, RMSE = 0.4720, and MAE = 0.3526, demonstrating excellent linear modeling capability. The RF model, assessed via 10-fold cross-validation, provided robust performance with R2 = 0.8819, RMSE = 0.6189, and MAE = 0.4653, effectively capturing potential non-linear relationships. These results highlight the effectiveness of reflectance-based spectral indices combined with multivariate regression as a reliable, rapid, and non-invasive tool for assessing sugar content in cashew apples. This methodology supports precision agriculture by enabling accurate in-situ quality evaluation of fruits.