Esmiedollar/digital_loans_financial_wellbeing_kenya

Python code and analysis notebooks for the MSc thesis: "Promise or Peril? County-Level Evidence on Digital Loans and Financial Well-Being in Kenya Using Machine Learning" Strathmore University, 2026. # Promise or Peril? County-Level Evidence on Digital Loans and Financial Well-Being in Kenya Using Machine Learning ## Overview This repository contains the complete Python code and Google Colab notebook supporting the MSc thesis submitted in partial fulfilment of the requirements for the degree of Master of Science in Mathematical Sciences and Risk Analytics at Strathmore University, Nairobi, Kenya (2026). **Author:** Esmie Awonteme Abaaluk **Submission Year:** 2026 **GitHub:** Esmiedollar --- ## Thesis Abstract The rapid expansion of digital lending in Kenya has transformed access to credit, yet its implications for household financial well-being remain contested. This study investigates whether digital loans represent promise or peril for Kenyan households using a rigorous risk analytics framework. Using nationally representative FinAccess survey data, the study models the probability of household financial distress as a function of digital loan usage and socioeconomic characteristics. A binary classification framework is adopted, beginning with survey-weighted logistic regression, followed by Random Forest and Gradient Boosting ensemble machine learning models. Model performance is evaluated using AUC ROC and Brier Score metrics. Predicted probabilities are aggregated to generate county-level financial vulnerability maps across all 47 Kenyan counties. SHAP explainability analysis is used to interpret model predictions. The findings confirm that digital loan usage is significantly associated with financial distress, with this association mediated by underlying socioeconomic characteristics including income, employment status, education, and health status. --- ## Repository Structure ``` digital_loans_financial_wellbeing_kenya/ │ ├── python_code_Esmie.ipynb ← Main analysis notebook └── README.md ``` --- ## Notebook Contents The single notebook `python_code_Esmie.ipynb` contains the complete end-to-end analysis, structured in the following section …