🌊 Flood Risk Prediction

A data-driven machine learning project to forecast flood risks in India using historical rainfall, elevation, and flood history datasets.
This project leverages multiple ML models to classify regions into High, Medium, and Low flood risk levels, enabling proactive disaster management and preparedness.

---

📌 Project Overview

Flooding is a major cause of damage to infrastructure, life, and the economy in India. With increasing climate unpredictability, accurate flood risk prediction is essential for timely evacuation and planning.

This project was developed as part of the Python for Data Science course, integrating rainfall, elevation, and flood history data, performing preprocessing and exploratory analysis, and applying machine learning models to predict flood vulnerability.

---

✨ Features

• Data preprocessing: handling missing values, outliers, and merging datasets.
• Exploratory Data Analysis (EDA): histograms, boxplots, correlation heatmaps, scatter plots, and trend analysis.
• Feature engineering: creation of a Flood Vulnerability Score.
• Machine Learning Models:
  • Random Forest
  • Linear Regression
  • Logistic Regression
  • Multi-Layer Perceptron (MLP)
• Model evaluation using MAE, RMSE, F1-score, and Variance Score.
• ROC curve and AUC analysis for model comparison.

---

📊 Datasets

1. Rainfall Data – Monthly and seasonal rainfall across Indian subdivisions (1901–2015).
   Source: Kaggle

2. Elevation Data – Elevation values (in meters) for Indian districts.
   Source: Kaggle

3. Flood History Data – Flood events across India (1967–2023).
   Source: Zenodo

4. Final Merged Dataset – Combination of rainfall, elevation, and flood history, including engineered vulnerability scores.

---

🛠️ Tech Stack

• Language: Python
• Libraries: pandas, NumPy, scikit-learn, fuzzywuzzy, matplotlib, seaborn
• Environment: Google Colab / Jupyter Notebook

---

⚙️ Installation & Setup

1. Clone this repository:

   git clone https://github.com/uncodingthecode/Flood-Risk-Prediction.git
   cd Flood-Risk-Prediction

2. Install dependencies:

   pip install -r requirements.txt

   If requirements.txt is missing, install manually:

    pip install pandas numpy scikit-learn matplotlib seaborn fuzzywuzzy

3. Open the notebook:

   jupyter notebook Flood_Risk.ipynb

or run it directly in Google Colab.

---

▶️ Usage

Run all cells in Flood_Risk.ipynb.

The notebook will:

• Load and preprocess datasets.
• Perform EDA (visualizations of rainfall, elevation, flood history).
• Train ML models on merged data.
• Evaluate performance using metrics & ROC curves.
• Output flood risk predictions (High/Medium/Low).

---

📈 Results

Model	MAE	F1 Score	RMSE	Variance Score
Random Forest	0.022	0.984	0.193	0.837
Linear Regression	0.263	0.987	0.382	0.979
Logistic Regression	0.088	0.917	0.296	0.661
MLP	0.083	0.982	0.112	0.998

✅ Best Performer: Multi-Layer Perceptron (MLP) with highest F1 score and lowest error values.

---

👨‍💻 Authors

• Naimish Shah
• Palash Shah
• Sk Qeyame Azam
• Aradya Shetty
• Aishani Singh

📌 Guided by Prof. Nikita Mishra

---

📜 License

This project is for academic purposes. License can be added based on future usage.