shixhellen/Alternative-Credit-Scoring-For-Kenya-s-Informal-Sector

The Alternative Credit Scoring System is a machine learning-based web application that assesses the creditworthiness of Kenya's informal population using alternative financial data such as M-Pesa usage, savings behaviour, and SACCO membership to support fair, transparent lending decisions. # Alternative Credit Scoring System for Kenya's Informal Sector ## Overview An explainable machine learning-based credit scoring system designed to assess the creditworthiness of financially underserved populations in Kenya using alternative financial behavioural indicators. Traditional credit scoring systems often exclude individuals without formal banking histories. This project explores how digital financial footprints, including mobile money usage, savings behaviour, and financial participation indicators, can support more inclusive lending decisions while maintaining transparency through Explainable AI (XAI). --- ## Problem Statement Millions of Kenyans participate actively in the digital economy but remain underserved by traditional lending systems due to limited formal credit histories. This project investigates whether alternative financial behaviour signals can predict credit risk and provide a fairer, more inclusive approach to credit assessment. --- ## Objectives The project aims to: - Develop a machine learning model for credit risk prediction using alternative financial indicators. - Explore the relationship between digital financial behaviour and repayment risk. - Build an explainable scoring framework to improve transparency in lending decisions. - Store and manage credit assessment records using a lightweight database system. --- # Project Features ## Data Processing and Exploration - Data cleaning and preprocessing pipelines. - Exploratory Data Analysis (EDA). - Feature engineering from financial behaviour indicators. - Statistical analysis of credit risk patterns. ## Machine Learning The project implements machine learning workflows including: - Model training and evaluation. - Credit risk classification. - Performance comparison between different models. - Model optimisation and selection. ## Explainable AI (XAI) Explainability techniques are used to: - Understand model predictions. - Identify important risk factors. - Improve …