Flood Prediction Using Machine Learning

Abstract

Flood prediction is of prime importance in management and mitigation of flood risks in flood-prone areas. In this project, a machine learning-based system has been proposed to predict floods on the basis of parameters like temperature, cloud cover, and humidity. From several fitted algorithms, such as Decision Tree, Random Forest, K-Nearest Neighbors, and XGBoost, the best predictive model was determined. XGBoost turned out to be the most precise, with an f1-score. In order to make this model practical, a Flask web application was created whereby users could input data and be given flood predictions easily. Indeed, the success of this system goes on to prove that through the use of machine learning, the basis of flood prediction can be highly improved to enable communities to adequately prepare for such an event. In order to improve this system, more historical data can be used with model parameter refinement. Thus, machine learning has a bright future in the sector of natural disaster management.

Introduction

???? Overview

Floods are increasingly devastating due to climate change, urbanization, and deforestation. Traditional flood prediction systems (e.g., HEC-RAS, SWAT) rely on physical models, which are region-specific, slow to adapt, and limited in handling real-time data. Machine Learning (ML) offers a more scalable and adaptive alternative.

???? Proposed System

The system uses supervised ML algorithms like Random Forest, Decision Tree, and K-Nearest Neighbors (KNN) to predict flood risk (high/medium/low) based on real-time and historical environmental data such as:

• Rainfall

• River water level

• Temperature

• Humidity

• Soil moisture

It’s integrated with IoT sensors, weather APIs, and supports real-time alerts.

? Key Advantages

• High Accuracy: Handles nonlinear patterns better than traditional models.

• Scalability: Works across regions with minimal recalibration.

• Real-Time Processing: Quickly reacts to new sensor or weather data.

• Adaptability: Continuously improves with more data.

• Early Warnings: Enables evacuation and disaster response planning.

• Cost-Effective & Automated: Minimal manual intervention after setup.

• User-Friendly Interface: Simple web-based input/output for users.

• Integration: Feeds into decision support systems for agencies.

????? System Architecture

1. Data Acquisition: From sources like IMD, Kaggle, IoT, and weather APIs.

2. Preprocessing: Handles missing data, normalization, label encoding, and data balancing using SMOTE.

3. ML Layer: Trains models (Random Forest, KNN, Decision Tree) on labeled data.

4. Prediction Layer: Outputs flood risk based on new inputs.

5. Frontend Interface: Built using Flask/Streamlit with visualization and risk display.

???? Algorithms & Techniques

• Random Forest: Most accurate; ensemble of decision trees.

• Decision Tree: Simple and interpretable.

• KNN: Pattern recognition based on data similarity.

• Supervised Learning: Uses labeled historical data.

• Cross-validation: Ensures model generalization and avoids overfitting.

????? Tools Used

• Python (backend & ML)

• Scikit-learn (ML algorithms)

• Pandas & NumPy (data manipulation)

• Matplotlib & Seaborn (visualization)

• Flask/Streamlit (frontend)

• GitHub (version control)

• SQLite/Firebase (optional data storage)

???? Results

• Random Forest achieved the highest prediction accuracy.

• Real-world datasets validated the model’s reliability.

• Web interface successfully displayed categorized flood risks and visual indicators.

Conclusion

The flood prediction system developed using machine learning has proven to be an effective tool for forecasting potential flood risks based on environmental factors like rainfall, river levels, humidity, and temperature[17]. By applying algorithms such as Random Forest and Decision Tree, the system can analyze large amounts of data and provide timely, accurate predictions[4]. This enables early warning, better planning, and faster emergency response in flood-prone areas. The integration of a user-friendly frontend ensures that both authorities and the general public can easily access and understand the predictions[20].

References

[1] Basha, E. A., &Rus, D. (2007). Design of early warning flood detection systems for developing countries. Proceedings of the 2007 International Conference on Information and Communication Technologies and Development. [2] Bhattacharya, B., &Solomatine, D. P. (2006). Machine learning in real-time hydrological forecasting. Hydrology and Earth System Sciences, 10(4), 789–805. [3] Chapi, K., et al. (2017). Flood hazard assessment using a novel ensemble decision tree-based model in GIS. Journal of Hydrology, 554, 704–716. [4] Chen, Y., et al. (2015). Application of artificial neural networks in flood forecasting. Journal of Hydrology, 529, 608–617. [5] El-Shafie, A., et al. (2011). Adaptive neuro-fuzzy inference system for rainfall forecasting in Klang River Basin, Malaysia. International Journal of Physical Sciences, 6(8), 1997–2003. [6] Jain, S. K., & Kumar, S. (2012). Flood hazard mapping using satellite images and GIS: A case study of Kosi River basin, India. Water Resources Management, 26(7), 2121–2134. [7] Kourgialas, N. N., &Karatzas, G. P. (2017). A flood risk decision support system for urban areas using GIS and artificial intelligence. Environmental Modelling & Software, 95, 12–21. [8] UNISDR. (2015). Global Assessment Report on Disaster Risk Reduction. United Nations Office for Disaster Risk Reduction. [9] Mosavi, A., et al. (2018). Flood prediction using machine learning models: Literature review. Sustainability, 10(11), 4200. [10] Roy, P. S., &Saha, S. K. (2019). Drought and flood monitoring using remote sensing. Indian Journal of Remote Sensing, 47(5), 707–715. [11] Sudheer, K. P., Gosain, A. K., &Ramasastri, K. S. (2002). A data-driven algorithm for nonlinear hydrologic system modeling. Hydrological Processes, 16(7), 1325–1330. [12] Yaseen, Z. M., et al. (2018). Artificial intelligence-based models for flood prediction: A review. Environmental Modelling & Software, 101, 124–132.GitHub.(n.d.).Flood Prediction Projects .Retrieve from https://github.com/search?q=flood+prediction+machine+learning [13] Basheer, A. K., et al. (2021). Real-time flood forecasting using artificial intelligence and satellite data. Water Resources Research, 57(9), e2021WR030193. [14] Zhang, Y., Wang, Y., & Li, X. (2019). Flood prediction using machine learning models: A review. Water, 11(12), 2513. [15] OpenWeatherMap API. (n.d.). Retrieved from https://openweathermap.org/api [16] Kaggle. (n.d.). Rain in Australia Dataset. Retrieved from https://www.kaggle.com/jsphyg/weather-dataset-rattle-package [17] Scikit-learn. (n.d.). Machine Learning in Python. Retrieved from https://scikit-learn.org/ [18] Pandas Development Team. (n.d.). Pandas Documentation. Retrieved from https://pandas.pydata.org/ [19] NOAA - National Oceanic and Atmospheric Administration. (n.d.). Retrieved from https://www.noaa.gov/ [20] Python Software Foundation. (n.d.). Python Official Documentation. Retrieved from https://docs.python.org/3/

Copyright

Copyright © 2025 Mugi Satish, Karri Harshitha. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.