Automated credit assessment framework using ETL process and machine learning

Area of Science:

• Financial Risk Management
• Machine Learning Applications
• Business Intelligence

Background:

• Real-time enterprise data analysis via Business Intelligence (BI) is vital for operational and strategic decisions.
• Automated ETL (extraction, transformation, load) processes enable near real-time data ingestion for BI.
• Automated credit decision-making systems enhance risk management, operational efficiency, and regulatory compliance for lenders.

Purpose of the Study:

• To develop and evaluate an automated credit risk assessment methodology for the financial domain.
• To leverage machine learning classification techniques for self-regulating data categorization in credit scoring.
• To integrate automated ETL processes for data preparation in machine learning model development.

Main Methods:

• Empirical approach using logistic regression and neural network classification.
• Adherence to Basel II standards for calculating expected loss.
• Implementation of an automated ETL process for data integration and machine learning model building.

Main Results:

• Demonstrated the feasibility of machine learning for automated credit risk assessment.
• Validated the integration of automated ETL processes for efficient data handling.
• Showcased compliance with Basel II standards in the proposed methodology.

Conclusions:

• The proposed machine learning-based methodology offers an effective approach to automated credit risk assessment.
• Automated ETL processes are crucial for enabling real-time data-driven credit decisions.
• The study provides a foundation for enhanced risk management and operational efficiency in financial institutions.