A comprehensive approach to flood detection using a hybrid artificial intelligence framework and geographic information system (GIS)

Mohamed, Mazlan and Siti Maesaroh and Afiyati, Afiyati (2025) A comprehensive approach to flood detection using a hybrid artificial intelligence framework and geographic information system (GIS). Borneo Journal Of Sciences And Technology (BJoST), 7 (2). pp. 99-111. ISSN 2672-7439

Text 028622310202512352353.pdf Download (1MB)

Abstract

This study implements the Random Forest model for a flood detection and early warning system by integrating rainfall and water level data. The dataset includes three levels of flood warning status, with a distribution of over 4,000 cases for statuses 1 and 2, and fewer than 500 cases for status 3. A scatter plot visualization shows a clear correlation between rainfall intensity and water level, where status 3 (the highest warning level) predominantly occurs at high rainfall intensity. The optimized Random Forest model, using GridSearchCV, demonstrates excellent performance with 100% accuracy for statuses 1 and 2, and 99% accuracy for status 3. The confusion matrix confirms the model's reliability, with only one misprediction out of 1,757 samples, where one status 3 cases was predicted as status 2. Decision boundary analysis reveals the model's capability to distinguish flood risk zones based on rainfall intensity and water level characteristics. These results demonstrate the effectiveness of a machine learning approach in developing an accurate and reliable flood early warning system.

Item Type:	Article
Uncontrolled Keywords:	Random Forest, Geographic Information System, Flood Detection, Machine Learning, Early warning system
Divisions:	Faculty of Artificial Intelligence and Cyber Security
Depositing User:	Norfaradilla Idayu Ab. Ghafar
Date Deposited:	15 Jul 2026 00:45
Last Modified:	15 Jul 2026 00:45
URI:	http://eprints.utem.edu.my/id/eprint/29939
Statistic Details:	View Download Statistic

Actions (login required)

View Item