Identifying Key Factors Causing Flooding Using Machine Learning
4
Issued Date
2025-01-01
Resource Type
eISSN
27236471
Scopus ID
2-s2.0-85216808018
Journal Title
Journal of Applied Data Sciences
Volume
6
Issue
1
Start Page
115
End Page
130
Rights Holder(s)
SCOPUS
Bibliographic Citation
Journal of Applied Data Sciences Vol.6 No.1 (2025) , 115-130
Suggested Citation
Gama A.W.O., Dennatan M., Dharmayasa I.G.N.P., Maw M.M., Sugiana I.P., Suryanti I. Identifying Key Factors Causing Flooding Using Machine Learning. Journal of Applied Data Sciences Vol.6 No.1 (2025) , 115-130. 130. doi:10.47738/jads.v6i1.463 Retrieved from: https://repository.li.mahidol.ac.th/handle/123456789/104199
Title
Identifying Key Factors Causing Flooding Using Machine Learning
Author's Affiliation
Corresponding Author(s)
Other Contributor(s)
Abstract
The impact of flooding extends beyond physical and infrastructural damage, affecting social, economic, and environmental dimensions. This study aims to identify the key factors influencing flooding by developing a decision tree model. The research method applies the C4.5 algorithm to build a decision tree model using flood factors such as rainfall, soil type, elevation, land use, and distance from rivers. The model is then applied to 57 past flood data events to determine key contributors to flooding in Denpasar City, Bali, Indonesia. The analysis showed that land elevation is the most influential factor, with areas below 28 meters above sea level having a 71% likelihood of being flood vulnerability. Additionally, the model reveals unknown patterns contributing to flood vulnerability among the factors considered. These insights give a deeper understanding of how these factors combine to affect flood vulnerability. The model's effectiveness was evaluated using a confusion matrix, resulting in an accuracy rate of 90%, a precision rate of 100%, a sensitivity rate of 90%, a specificity rate of 100%, and a F1 Score rate of 94%, demonstrating its strong predictive power in identifying areas at risk of flood vulnerability. Although this study is limited by the availability of data, the focus on Denpasar City, and the potential omission of other relevant attributes, it advances flood risk assessment by applying machine learning to provide practical insights that could enhance flood management strategies, with potential applications to other urban areas facing similar risks.