Analysis of Forecasting Methods on Rice Price Data at Milling Level According to Quality

Rice is a primary source of carbohydrates for many Indonesians, and its prices often surge due to uncontrolled demand. Therefore, the government is crucial in monitoring rice prices to maintain stability. Information technology, particularly data mining such as forecasting, is essential for providing accurate information on future rice prices. It will assist various stakeholders in making informed pricing policy decisions. This study employs Random Forest Regression and Gradient Boosting Regressor methods to predict rice prices using a dataset that includes monthly average rice prices at milling levels, categorized by quality (Premium and Medium), spanning from January 2013 to April 2024. The dataset consists of 136 rows, each representing a unique combination of year, month, and quality, and is stored in CSV format. Methodological steps include data collection, preprocessing, modeling, and model evaluation using monthly average rice prices at milling levels based on quality, including premium and medium grades. The results from Random Forest Regression indicate Root Mean Square Error (RMSE) values of 24.90 for premium rice and 25.47 for medium rice. The study reveals that Random Forest Regression outperforms Gradient Boosting Regressor in this context. Future research should explore additional prediction methods and consider other variables influencing rice prices to enhance model accuracy.