AI (BNDE)

PROBLEM

The challenge in the business loan arena is to prevent back-and-forth communication between the risk managers, account managers and the back office. The aim is to reduce the turnaround time.  

APPROACH

Over the past five years at BNDE, we accumulated data about back-and-forth communication between the loan applications and risk departments. 

Instead of predicting the probability of default based on the risk models, we sought to predict the probability of back-and-forth communication between the risk and business departments. We implemented automatic data capture. We developed models that weight the factors according to their predictive value. A rule engine automatically detects a deficiency in credit preparation and notifies the sales representative.

RESULTS

The project is currently underway.  The model is able to predict with 90% accuracy the probability that a credit request will be returned for revisions and allows the rule weighting to be updated on an ongoing basis

EXPERTISE USED 

• DATA SCIENCE
• MACHINE LEARNING
• INFORMATION TECHNOLOGY