Predicting Flood Risk in Manado City Using the C4.5 Decision Tree Algorithm: A Data-Driven Approach

Apriandy Angdresey; Indah Yessi Kairupan; Ranodeyansa Rumanjar; Andre Gabriel Mongkareng; Ignatius Lucky Henokh Tangka

Apriandy Angdresey Department of Informatics Engineering, Universitas Katolik De La Salle, Manado.
Indah Yessi Kairupan Department of Information Systems, Institut Agama Kristen Negeri Manado, Indonesia.
Ranodeyansa Rumanjar Department of Informatics Engineering, Universitas Katolik De La Salle, Manado.
Andre Gabriel Mongkareng Department of Informatics Engineering, Universitas Katolik De La Salle, Manado.
Ignatius Lucky Henokh Tangka Department of Informatics Engineering, Universitas Katolik De La Salle, Manado.

Abstract

Flooding, a natural disaster commonly triggered by heavy rainfall or blocked waterways, poses a persistent threat to Manado City due to its proximity to the sea and major rivers, Tondano and Tikala. This study develops a flood prediction application using the C4.5 Decision Tree algorithm based on historical data (2017–2021) from five rivers. Input variables include rainfall, water discharge, runoff coefficients, and river cross-sectional data. The model supports the Sulawesi River Regional Office I by predicting flood conditions—flood, prone to flooding, or no flooding—to enhance mitigation strategies. Experimental results show robust predictive performance, achieving accuracies of 86.13%-87.07% across different data splits. Future integration with the Internet of Things (IoT) is proposed to enable real-time data acquisition, thereby improving the system's responsiveness and flood risk management effectiveness.