Rainfall intensity classification in Subang: a decision tree approach

Subang, Malaysia, is increasingly vulnerable to flash floods due to its urban development and frequent weather disturbances. Accurate rainfall forecasting is therefore essential for mitigating flood risks and supporting effective urban planning. This study employs Decision Tree Analysis to classify rainfall intensity into two categories: rain and no rain. Historical meteorological data, including temperature, humidity, wind speed, and mean sea level pressure, were analyzed to identify the key environmental determinants of rainfall. Three decision tree models, based on Gini, Entropy, and Logworth indices were applied to classify rainfall events and evaluate predictive performance. The analysis reveals that temperature and relative humidity exert the strongest influence on rainfall intensity, followed by mean sea level pressure. Among the models, the Gini index demonstrated superior accuracy in detecting rainfall events, whereas the Entropy-based model provided the most consistent performance in terms of generalization and sensitivity to unseen data. These findings highlight the capacity of decision tree methods to capture nonlinear interactions within meteorological variables while producing interpretable and practical forecasts. The study underscores the potential of such models to enhance flood management strategies and urban resilience. Furthermore, it recommends continuous data enrichment and the integration of additional meteorological variables to improve prediction accuracy.