Comparative Evaluation of Machine Learning Models for Pluvial Flood Susceptibility Assessment: A Case Study of Ibadan

Abstract

Pluvial flooding is a major urban hazard that occurs when intense rainfall exceeds the capacity of drainage systems, leading to surface water accumulation and disruption to urban activities. This paper examines pluvial flood susceptibility in Ibadan metropolis, Nigeria, using GIS-based data and a machine learning approach to evaluate flood risk levels. The study classifies flood susceptibility into five categories: Very High, High, Moderate, Low, and No Flood, based on flood-conditioning factors related to location, terrain, runoff behavior, drainage conditions, and rainfall. The data were divided into 80% for training and 20% for testing, and multiple machine learning models were evaluated within a consistent experimental framework using accuracy, macro and weighted precision, recall, F1-score, and confusion matrix analysis. The results show that all models achieved strong predictive performance, with Random Forest achieving the highest accuracy of 97.76% and the most consistent performance across classes. Most classification errors occurred between adjacent categories, such as Very High and High, indicating that uncertainty is mainly concentrated near class boundaries. This pattern reflects the gradual transition of environmental conditions, where neighboring susceptibility classes may share similar characteristics. The resulting five-class flood susceptibility map provides a practical tool for supporting drainage management, urban planning, and emergency preparedness in flood-prone areas.