Predictions of oil volume in palm fruit and estimates of their ripeness: A comparative study of machine learning algorithms

Recent advances in precision agriculture highlight the crucial role of machine learning in predicting crop yields by discerning intricate patterns in agro-meteorological data. However, its adoption in the oil palm industry in Thailand remains limited. This study aimed to compare machine learning algorithms for predicting the oil content from different parts of both ripe and raw oil palm fruits (top, middle, and down). Additionally, we compared algorithms for predicting oil volume in semi-ripe and unripe fruits. Among the methods used, Random Forest and Gradient Boosting models mostly excelled in predicting the oil content at different positions on the oil palm fruit. In contrast, Decision Trees and XGBoost were the most accurate predictors of oil volume for semi-ripe and unripe oil palm fruits, respectively. Overall, this research emphasizes the potential of machine learning to enhance oil palm industry practices and optimize agricultural strategies in Thailand.