Daily Rainfall Prediction based on Gradient Boosting Regression Model using NEX-GDDP-CMIP6

Rainfall prediction is crucial in guiding climate adaptation and mitigation strategies, particularly in regions vulnerable to climate variability like Indonesia. This study presents a machine learning approach for developing a rainfall prediction model using the NEX-GDDP-CMIP6 dataset from the EC-Earth3 model. Gradient Boosting Decision Trees (GBDT) was used to predict daily rainfall prediction due to their robustness, flexibility, and interpretability. A total of 5,372,741 samples from the Indonesian region, representing its climate conditions and atmospheric states, were utilized, with five climate-related features serving as input variables. The model's performance assessment has been conducted using evaluation metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R-squared (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>). Results demonstrated that Gradient Boosting model outperformed other tree-based methods, achieving an RMSE of 0.02 and an R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of 0.99, indicating high predictive accuracy. The findings suggest that GBDT is well-suited for daily rainfall prediction tasks using climate model datasets, offering a reliable tool for enhancing climate decision-making processes.