Evaluating Rainfall Prediction Models: A Comprehensive Analysis of Linear Regression, Gradient Boosting, and Random Forest in Multi-Location Data Sets

Rainfall prediction is a critical task for agricultural productivity and water resource management. This study aims to compare the performance of three machine learning models—Linear Regression, Gradient Boosting, and Random Forest—in predicting rainfall. The dataset, sourced from various meteorological offices (only rainfall value from Rorotan, tanjung Priok, rawa Badak, Kodamar, and Pulo gadung). The models were evaluated using metrics such as Root Mean Squared Error (RMSE) and R-squared ($\mathbf{R}^{2}$). Our findings indicate that Gradient Boosting outperforms both Linear Regression and Random Forest in terms of prediction accuracy. Specifically, Gradient Boosting demonstrated superior performance in handling the non-linear relationships and complex interactions among the atmospheric variables. Random Forest also showed robust performance but was slightly less accurate than Gradient Boosting. Linear Regression, while simpler and faster, lagged behind the other two models in predictive accuracy. These results suggest that advanced machine learning techniques like Gradient Boosting are more effective for rainfall prediction, providing valuable insights for agricultural planning and water resource management. These findings have implications for optimizing short- and long-term rainfall predictions.