Solving financial exclusion in Kenya through machine learning and spatial analysis to pinpoint "Financial Deserts" to maximize financial inclusion.
# π Financial Deserts: Strategic Expansion in Kenya
### π Project Roadmap
Structure of this ReadMe file:
1. **The Problem & Strategic Solution:** An overview of the economic friction caused by "Financial Deserts" and the market-capture solution.
2. **Deep Analysis & Methodology:** A technical summary of the data engineering and machine learning logic used.
3. **Interactive Visualizations:** Access to the hosted web application/dashboard.
4. **Running the Project:** Technical requirements and installation steps to replicate the analysis.
5. **Video Demonstration:** A video walkthrough located at the very bottom of this file.
---
β¨ **Technologies**
* **Python:** GeoPandas, Pandas, NumPy, Scikit-Learn
* **Spatial Analysis:** Shapely, EPSG:21037 Projections
* **Front-End:** React / Framer Motion / Power BI
π **The Problem & Strategic Solution**
* **The Problem: The "Financial Desert" Crisis.** While cities are crowded with bank agents, millions of people in rural Kenya live in "Financial Deserts". These are areas where the nearest financial hub is so far away that it traps entire settlements in isolation. Because roads are often poor or non-existent, a simple trip to withdraw money can require an exhausting journey on foot.
* **The Insight: The Opportunity Cost of Time.** In these areas, the true barrier is the **Opportunity Cost of Time**. When a community must spend several hours just to reach an agent, they lose valuable time that could be spent on farming, labor, or education. Every hour spent walking is an hour of lost economic productivity for the whole region.
* **The Solution: Market-Capture through Expansion.** This project treats agent expansion as a **market-capture strategy**, not just a social initiative. By reducing the travel burden, we "unlock" the dormant economic potential of these settlements.
* **The Analytical Approach: Engineering an "Impact Score."** I analyzed 9,900+ towns against 1,737 financial service points. By projecting the map into β¦