Improving Operation Efficiency through Predicting Credit Card Application Turnaround Time with Index-based Encoding

Abstract

This paper presents the successful use of index encoding and machine learning to predict the turnaround time of a complex business process - the credit card application process. Predictions are made on in-progress processes and refreshed when new information is available. The business process is complex, with each individual instance having different steps, sequence, and length. For instances predicted to have higher than normal turnaround time, model explain-ability is employed to identify the top reasons. This allows for intervention in the process to potentially reduce turnaround time before completion.