<![CDATA[Tomato production in each regency/municipality changes from year to year, necessitating a predictive method that can support production planning, distribution, and agricultural policy decision-making. This study aims to develop and compare machine learning models for predicting tomato production based on historical data from regencies/municipalities in Aceh Province. The dataset consists of 230 observations from 23 regencies/municipalities covering the period from 2015 to 2024, with attributes including region, year, and tomato production measured in quintals. The research stages include data cleaning, handling missing values, data transformation, and the construction of historical features, including production in the previous year, the three-year average production, and the production growth rate. The algorithms evaluated include Linear Regression, Random Forest Regressor, XGBoost Regressor, and CatBoost Regressor. The data are divided chronologically to prevent data leakage, while model performance is evaluated using Mean Absolute Error, Root Mean Squared Error, and the coefficient of determination. Preliminary analysis indicates that the production data have a very wide range and an uneven distribution pattern, requiring logarithmic transformation. This study is expected to identify a predictive model with the lowest error rate and provide supporting information for local governments and stakeholders in planning tomato commodity development in a more measurable and systematic manner.]]>
