A Data-Driven Approach for Extracting and Characterizing Chemical Attributes of Mango Fruit

Mango fruit is widely recognized for its substantial nutritional value and its contribution to a balanced diet. It is an excellent source of vitamin C, vitamin A, and dietary fiber that facilitates digestive function. This fruit also provides many important minerals, including potassium and magnesium, and it is rich in bioactive compounds, particularly polyphenols. It is an important tropical fruit crop widely cultivated in Saudi Arabia, especially in the southwestern regions, where climatic conditions are favorable for its production. In this paper, we propose novel techniques for analyzing the nutritional composition and quality attributes of mango. The approach integrates spectroscopic techniques with advanced statistical modeling to evaluate and predict the quality attributes of mango. Specifically, we employ some functional regression methods, including least squares relative error regression(LSRE), least absolute relative error regression (LARE), and functional least absolute deviations regression (FLAD), to predict important indicators such as vitamin C content, soluble solids content (SSC), and total acidity (TA). The results demonstrate that these functional approaches outperform conventional methods, such as partial least squares regression (PLSR) and principal component regression (PCR), in terms of predictive accuracy. Furthermore, the proposed models exhibit strong robustness, which maintains reliable performance under heterogeneous and non-normal data conditions, highlighting their suitability for practical applications in fruit quality assessment.