Innovative Flood and Landslide Prediction using Machine Learning

Abstract

Natural disasters such as floods and landslides pose serious threats to human life, infrastructure, and the 
environment, particularly in regions with complex geographical conditions and vulnerable populations. This study 
presents a machine learning–based framework for the prediction of floods and landslides using historical weather 
data, rainfall records, soil characteristics, and other environmental parameters. Several machine learning 
algorithms, including Logistic Regression, Random Forest, Support Vector Machine (SVM), K-Nearest Neighbors 
(KNN), Gradient Boosting, and XGBoost, were developed and evaluated to identify the most effective prediction 
models. Performance evaluation was carried out using accuracy, precision, recall, and F1-score metrics. Among 
the tested models, Random Forest achieved the best performance for flood prediction, while Gradient Boosting 
provided superior results for landslide prediction, with both models attaining an accuracy of nearly 97% on the test 
dataset. Exploratory Data Analysis (EDA) was conducted to analyze data distribution, identify feature correlations, 
and detect outliers, thereby improving model efficiency and reliability. The proposed system was implemented as a 
Flask-based web application that enables real-time prediction, visualization, and model comparison. The results 
demonstrate the effectiveness of machine learning techniques in supporting disaster prediction and early warning 
systems.