Home Credit Group grants loans to people most banks declined. Thus the company relies on keeping an uneasy balance between granting a loan to high-risk individuals and never recouping its investment.
To spot the difference between customers who are and who are not willing or able to repay debt, the company combines data from multiple sources, validating it and uncovering the patterns.
By applying deep learning solutions and deep data exploration, deepsense.ai’s team delivered a model that spotted, with an 80% success rate, payers in the riskiest group who were unwilling or unable to repay the debt.
The system was delivered as a part of a competition co-organized by Home Credit Group and the Google-owned Kaggle platform.