Decision Tree-Based Weather Factor Analysis for Bandar Lampung Flood Event Classification

Abstract

Flood events in Bandar Lampung City, Indonesia, are often triggered by extreme weather factors such as high rainfall, high humidity, and fluctuating air temperatures. This study applies the Decision Tree model as an analytical tool to examine the influence of these weather factors on flood occurrences using a dataset of 50 historical records containing information on rainfall, air temperature, humidity, and wind direction. The methodology includes data exploration, preprocessing, a single train–test split (80:20), and model evaluation using a confusion matrix. The exploratory analysis revealed that daily and weekly rainfall are the dominant factors contributing to flood events. The Decision Tree model, trained using default parameters, achieved a prediction accuracy of 90% on the test data. The visualization of the decision tree and feature importance analysis reinforce the finding that rainfall is the primary predictor of flood events. As a preliminary proof of concept, this study demonstrates that the Decision Tree model provides an interpretable approach for flood prediction. However, given the limited dataset size, further validation using larger datasets is recommended before operational implementation in early warning systems for Bandar Lampung City.