Credit Scoring Analysis & Model for People Without Traditional Banking

This project involved developing a credit scoring system for individuals lacking access to traditional banking, leveraging alternative data sources like mobile phone usage, utility payments, and community-based lending behavior. The model used Random Forest to predict creditworthiness, achieving an accuracy of 60%. The analysis highlighted the potential for financial inclusion for traditionally underserved populations.

Key Insights:

  • Mobile Phone Payment Behavior: Identified as a key predictor of creditworthiness, with a 65% correlation between timely mobile payments and higher credit scores.
  • Community-Based Lending Behavior: Found to have a 50% predictive power for assessing loan repayment capabilities, highlighting the importance of informal lending practices in determining financial reliability.
  • Default Prediction Rate Reduction: The model successfully reduced the default prediction rate by 15% compared to traditional credit scoring methods, identifying lower-risk individuals who may have been overlooked.

Project Impact:

  • Financial Inclusion: Provided an alternative credit scoring system that enables previously excluded individuals to access financial services such as loans and credit lines, promoting economic participation.
  • Data-Driven Decisions: Empowered financial institutions to make more informed decisions, improving loan approval processes for underrepresented populations.
  • Policy Implications: Highlighted the importance of considering non-traditional financial data in regulatory frameworks, potentially influencing policy reforms to increase financial inclusion in underserved communities.