<![CDATA[Food security was an important issue influenced by production, access, prices, and socio-economic conditions. In Indonesia, the Food Security Index (IKP) was used as the main indicator. However, prediction methods such as multiple linear regression often failed to capture the complex relationships between variables. Machine learning methods, such as random forest regression, offered a more suitable alternative for non-linear and large-scale data. Nevertheless, few studies in Indonesia compared the effectiveness of these two methods. Therefore, this study aimed to compare the performance of linear regression and random forest in predicting the IKP, in order to support more accurate and sustainable food security planning. The analysis results showed that the forecasting method with better performance in predicting the IKP in Indonesia was Random Forest Regression. This study made a significant contribution by empirically comparing multiple regression and Random Forest in predicting the Food Security Index (IKP) using big data. The results showed that Random Forest performed better in terms of MSE (5.5431) and RMSE (57.7242), indicating higher overall accuracy, while multiple regression had lower MAE (6.0805) and slightly higher R² (68.21%), suggesting more stable predictions and better explanatory power. Random Forest also identified key influencing variables, such as poverty rate and health worker ratio, and provided clearer insights through decision tree visualization. Overall, the findings demonstrated that while no model was entirely dominant, Random Forest offered greater flexibility and predictive strength for complex, large-scale data, supporting its potential use in formulating data-driven food security policies in Indonesia]]>
Copyrights © 2025
