A machine learning system that leverages GIS data to predict flood risk areas in Kenya
# 🌍 Flood Risk Prediction System (Kenya)
## 📌 Business Understanding
Flooding is a major environmental and socio-economic challenge in Kenya, impacting infrastructure, agriculture, and human safety. Traditional flood risk assessment methods often lack integration of diverse geospatial factors.
In this project, I built a machine learning-driven system that leverages GIS data to predict flood risk, enabling stakeholders such as urban planners, disaster response teams, and policymakers to make informed decisions.
---
## đź“– Project Overview
This project integrates **Geographic Information Systems (GIS)** and **Machine Learning** to assess flood risk across Kenya.
I implemented two modeling approaches:
- **Classification** → Predict flood risk categories (Low, Moderate, High)
- **Regression** → Predict a continuous flood risk score
The dataset was engineered from multiple geospatial data sources using:
- `GeoPandas`
- `Rasterio`
- `WhiteboxTools`
---
## âť— Problem Statement
Flood risk prediction is complex due to the interaction of environmental, topographic, and human-related factors.
This project aims to:
- Quantify flood risk using a numerical score
- Categorize regions into actionable risk levels
- Leverage spatial data for predictive modeling
---
## 🎯 Objectives
- Develop a **Flood Risk Score**
- Classify areas into **risk categories**
- Compare **classification vs regression approaches**
- Improve model performance using **ensemble techniques (Voting & Stacking)**
- Deploy a working prediction system
---
## 📊 Metrics of Success
### Classification
- Accuracy (75%)
- F1 Score (75%)
- Confusion Matrix
### Regression
- R² Score (70%)
- Mean Absolute Error (MAE) (0.5)
- Mean Squared Error (MSE) (0.5)
- Residual Analysis (Residual Plots)
---
## đź“‚ Data Understanding
### 📌 Dataset Features
1. **county**
- The administrative region in Kenya where the data point is located. This categorical feature enables geographic grouping and regional analysis of flo …