Comparing the Performance of Three Decision Tree Models for Precipitation Prediction in Cengkareng Soekarno Hatta

Abstract

This study evaluates and compares the performances of advanced decision tree algorithms, namely LightGBM, XGBoost, and Catboost, for precipitation prediction in Cengkareng Soekarno Hatta, a region in Jakarta, Indonesia that experiences heavy rainfall and flooding. Using historical rainfall data, the study evaluates the algorithm based on various metrics such as accuracy, precision, recall, F1 score, and ROC/AUC. The study aims to provide insight into the potential of advanced machine learning algorithms for precipitation prediction in the region. This study illustrates various visualization and analysis techniques for exploring and understanding data. The analysis employs three decision tree models, including LightGBM, XGBoost, and Catboost, to predict rain patterns. Results indicate that all models perform well in predicting no rain, but struggle with moderate and heavy rain. AUC values for all models are similar, indicating their equal ability to distinguish between positive and negative instances. The study highlights the importance of using appropriate visualization techniques to gain insights into the data and explores the correlations between different features to better understand the underlying relationships.