2OS built a new advanced machine learning model for our client. It was the first time an advanced machine learning model was used at this issuer. The model significantly outperformed their current logistic regression model.
The Best for Every Customer
Existing customer should be treated according to their needs. The Machine Learning models helped the bank to reward the strongest customers with extra line and to protect the most risky customers with line decreases.
The 2OS model allowed our client to better identify potential risk in their portfolio. The client improved risk separation for the best 10% customers by 74%, while also improving risk separation for the worst 10% of customers by 18%.
The 2OS model will be used to improve the profitability of the issuer’s Credit Line Increase Program, Credit Line Decrease Program, Authorization Restriction, and Balance Transfer Program.
Prediction Improvement in the top 10% of customers
Prediction Improvement in the bottom 10% of customers