Analysis of KNN and Decision Tree for Rainfall Prediction in South Tangerang

Abstract

The phenomenon of variable rainfall and extreme rain with high intensity occurring within a short period frequently takes place in South Tangerang City and disrupts community activities. The uncertainty of daily rainfall patterns poses a serious challenge in disaster mitigation and regional planning. This condition requires a scientific approach based on historical data capable of accurately identifying weather patterns. However, comparative research between two data mining algorithms, namely K-Nearest Neighbor and Decision Tree C4.5, for rainfall prediction in South Tangerang City is still limited, thus a more in-depth analysis of their performance is necessary. This study aims to analyze and compare the performance of K-Nearest Neighbor and Decision Tree C4.5 in rainfall prediction using weather classification categories in South Tangerang City. The data were obtained from BMKG for the period 2022–2024 and processed using RapidMiner software through several stages, including preprocessing, data cleaning, normalization, data splitting, model evaluation, and testing using 5-fold and 10-fold cross-validation. The evaluation metrics used were accuracy, recall, precision, F1-score, and kappa.The expected outcome of this study is to obtain a classification algorithm capable of predicting rainfall accurately and consistently based on historical weather patterns. Based on the testing results, Decision Tree C4.5 demonstrates a higher accuracy level compared to K-Nearest Neighbor, thus it can serve as a scientific foundation for the South Tangerang City Government for further development.