Application of machine learning for short-term climate prediction in Indonesia

Abstract

This study explores the Application of Machine Learning for Short-Term Climate Prediction in Indonesia, focusing on enhancing forecast accuracy through advanced computational models. The primary objective was to develop and validate Random Forest and Support Vector Machine (SVM) models to predict short-term climate conditions accurately across ten major Indonesian cities. Employing a quantitative approach, the study utilized experimental design, rigorous data analysis, and model validation using historical weather data from April 2024 provided by the Indonesian Meteorological, Climatological, and Geophysical Agency (BMKG). The results indicate that both Random Forest and SVM significantly outperform traditional climate prediction models, with Random Forest achieving an average accuracy of 87.5% and SVM 85.2%. These findings underscore the potential of machine learning to revolutionize short-term climate predictions in regions with complex meteorological dynamics like Indonesia, offering substantial implications for disaster preparedness, agricultural planning, and urban management. Future research can expand upon these models by incorporating real-time data and exploring deep learning techniques to enhance predictive reliability further