Assessing Machine Learning Algorithms in Sablayan Occidental Mindoro for Data-Driven Rice Yield Prediction

Predicting rice yields accurately is essential for maintaining food security, allocating resources as efficiently as possible, and promoting sustainable farming methods. This study assesses the performance of four machine learning algorithms Random Forest, Naïve Bayes, Logistic Regression, and KStar using a dataset of 180 instances with 11 attributes. WEKA (Waikato Environment for Knowledge Analysis) with 10-fold cross-validation was used to develop and evaluate the models. Confusion matrices, precision, recall, F1 score, overall accuracy, and Kappa statistics were used to evaluate performance. Confusion matrices, precision, recall, F1 score, overall accuracy, and Kappa statistics were used to evaluate performance. The results showed that Random Forest outperformed all other algorithms, achieving the highest accuracy (99.44%) with a Kappa statistic of 0.957. In both classes, it showed excellent precision, recall, and F1 scores. The minority "linear" class, on the other hand, was difficult for Naïve Bayes and KStar to handle, while Logistic Regression did reasonably well but fell short of Random Forest. These results demonstrate Random Forest's sensitivity to misclassification errors and validate its effectiveness in predicting rice yield.