Accurate prediction of food consumption is essential for strengthening regional food security planning, particularly in areas experiencing increasing food demand and environmental uncertainty. This study aims to predict food consumption patterns in Sukabumi Regency, West Java, Indonesia, using an integrated artificial intelligence approach. The research combines the Adaptive Neuro-Fuzzy Inference System (ANFIS) for forecasting food consumption trends with three machine learning classification algorithms—Support Vector Machine (SVM), Random Forest (RF), and Logistic Regression (LR)—to classify food consumption levels. Historical rice consumption data from 2014 to 2024 were used to train the forecasting model and generate projections up to 2030. The ANFIS training process was conducted with 100 epochs and an error tolerance of 0, resulting in a training error value of 0.182, indicating strong model learning capability. The comparison between predicted and actual consumption values showed a prediction accuracy of 95.2%, demonstrating the reliability of the model in capturing consumption patterns. Furthermore, food consumption levels were classified into three categories: low, medium, and high. The classification results revealed that Random Forest achieved the most consistent performance across cross-validation folds, while SVM and Logistic Regression experienced misclassification in the medium consumption category. In several evaluation scenarios, machine learning models achieved accuracy levels up to 99.75%, precision 99.76%, recall 99.75%, and F1-score 99.75%. The integration of ANFIS forecasting and machine learning classification provides a robust analytical framework for understanding food consumption dynamics and supports data-driven policy formulation aimed at strengthening regional food security in Sukabumi Regency.
Sukmawani et al. (Tue,) studied this question.