theBrianMwenda/kenya-credit-risk-profiler

# 📊 Kenya Mobile Credit Risk Profiler > An end-to-end data analytics project segmenting mobile borrower risk profiles among Nairobi residents — combining behavioral finance theory with machine learning. --- ## 🌍 Project Context Kenya's mobile lending market is one of the fastest-growing in Africa — yet **over-indebtedness and default rates** remain high. Most credit risk models used by Kenyan fintechs were trained on Western data and don't reflect local borrower behavior. This project uses **primary survey data from 286 Nairobi CBD respondents** to build a Kenya-specific mobile credit risk segmentation model, grounded in behavioral finance theory and validated with Stata regression analysis. **This project is directly connected to my undergraduate research thesis** on consumer perception of mobile loan apps and personal debt management. --- ## 🔬 Research Framework The analysis applies three established theoretical frameworks: | Framework | Application in This Project | |-----------|----------------------------| | **Technology Acceptance Model (TAM)** | Perceived usefulness & ease of use as adoption predictors | | **Theory of Planned Behavior (TPB)** | Subjective norms and behavioral intention to borrow | | **Behavioral Finance Theory** | Loss aversion, mental accounting, and debt rationalization | --- ## 📁 Project Structure ``` kenya-credit-risk-profiler/ ├── data/ │ ├── raw/ # Original survey data (anonymized) │ ├── processed/ # Cleaned, analysis-ready datasets │ └── external/ # Supplementary Kenyan fintech data ├── notebooks/ │ ├── 01_data_cleaning.ipynb # Import, clean, validate survey data │ ├── 02_exploratory_analysis.ipynb # EDA — distributions, correlations, viz │ ├── 03_feature_engineering.ipynb # TAM/TPB score construction │ ├── 04_segmentation.ipynb # K-Means borrower clustering │ ├── 05_risk_classification.ipynb # Logistic Regression + Decision Tree │ …