Modeling Flood Prediction Using Logistic Regression Enhanced with Machine Learning

Abstract

Flooding remains one of the most destructive natural disasters in Nigeria, particularly in Lokoja, Kogi State, due to its location at the confluence of the Niger and Benue rivers. This study develops a data-driven flood prediction model by integrating logistic regression with machine learning techniques to improve early warning systems. Meteorological data, including rainfall, temperature, and relative humidity, were obtained from the Nigerian Meteorological Agency. Data preprocessing, exploratory analysis, and feature engineering were conducted using Python. The dataset (120 records) was split into training (80%) and testing (20%) sets. The model was evaluated using accuracy, precision, recall, and F1-score. Results show high predictive performance with 98.3% accuracy, 100% precision, 90.9% recall, and 95.2% F1-score. The model was deployed as a web-based application using Streamlit for real-time predictions. The findings demonstrate that combining statistical modeling with machine learning improves flood prediction reliability and supports disaster management decision-making.