Online investment-linked financial company Eight Percent announced on July 28 that it will apply its proprietary artificial intelligence (AI) credit scoring model, 'E-index 5.0,' to its personal credit loan products.
E-index 5.0 was developed based on approximately 7.2 million personal credit assessment data. It is designed to evaluate repayment capabilities more accurately by learning from various patterns, including approved customers, those rejected during the assessment process, early repayment customers, and those who have restored their credit after default.
The new model reduces the number of variables needed for assessment by about 70% compared to the previous model (E-index 4.0), while maintaining predictive performance, thereby enhancing assessment speed and stability.
The model also improves discrimination for low- and mid-credit borrowers. Internal simulations showed that the repayment capacity discrimination for these borrowers improved by 21.3% compared to using only traditional credit scoring, while overall customer discrimination improved by 8.2%.
This advancement reflects the incorporation of extensive assessment data into AI-based modeling to better evaluate the repayment capabilities of customers with limited financial transaction histories or diverse income sources, such as recent graduates, freelancers, and gig workers.
Eight Percent plans to use E-index 5.0 as a key infrastructure to expand linked investments for institutional investors in the future.
Park Kwon-soo, head of the CSS team at Eight Percent, stated, "This credit assessment model is the result of enhancing an AI-based machine learning model to select key signals necessary for determining repayment capability. In the future, we expect to improve the speed and stability of large-scale automated assessments and handle large-scale mid-interest loans more precisely when institutional investor funds flow in."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.