With businesses getting an overwhelming response, customer debts started to grow exponentially, and associated recovery was always a challenge at an overall level. The company had been storing a large volume of relationship data points for a long time, but none of this could be put to use to calculate the risk score. This resulted in the collections department’s is being handicapped in the recovery of certain bad debts. The data warehousing team at this big financial services company analysed large sets of customer records collected from sources such as loan management systems and credit bureau data.


The data warehouse architecture entailed importing all data into an OLAP database, calculating and speculating on a “Risk Profile” with human intervention, and then loading it into the ultimate master database.

How did Whiteklay overcome the challenges?

The IZAC analytical framework deployed on the Hadoop architecture revolutionised the building of a customer risk profiling use case with just a 15-node distributed computing setup. The solution utilised the power of Hadoop over Spark computing and machine learning processing frameworks to evolve a system that could auto-profile a customer and calculate a risk score associated with the same.