Machine Learning for Ethical Credit Scoring System

Background

Over 1.7 billion adults globally lack access to financial services, with a major barrier being the absence of a credit history or collateral required by traditional credit scoring systems. This leaves a large population excluded from borrowing opportunities, particularly first-time borrowers.

Objective

The goal of the project was to create an ethical credit scoring system aimed at banking the unbanked, using machine learning techniques to develop fair, transparent, and consumer-owned credit scores. This solution was particularly targeted at low-income individuals who are often overlooked by traditional credit scoring methods.

Approach

To tackle the challenge, the team collaborated with the AI startup Creedix, which is on a mission to develop the world’s #1 ethical credit scoring solution. The project utilized three key datasets provided by Creedix:

• Transaction data (account numbers, region, mode of transaction, etc.)
• Per capita income per area
• Job titles of account holders (all anonymous and privacy law compliant).

The team engineered features from this data, incorporating external sources like Numbeo (for cost of living data) and Indeed (for salary information). This allowed them to fill in gaps and enrich the features, creating a more accurate and robust dataset.

Results and Impact

The team successfully engineered a variety of features and developed machine learning pipelines for both supervised and unsupervised learning models. While the final model development and feature weighting were beyond the scope of this project, the deliverables significantly advanced the development of a fair and transparent credit scoring system. This project has the potential to promote financial inclusion by enabling ethical credit scoring for low-income individuals and first-time borrowers.

Future Implications

The ethical credit scoring system developed in this project could pave the way for future policies and research into more inclusive financial models. It provides a foundation for the further development of consumer-owned, transparent credit scoring systems that could positively influence global financial inclusion and provide access to credit for previously underserved populations.