Case Study 11 – How a manager used Analytics to Strengthen Fraud Detection Strategy during Credit Card Approval process.
Industry – Banking and Financial Services
We follow DCOVA and I methodology to solve the problem. To Understand this methodology, check this whitepaper
Business Problem – The manager has historical data of its customers who are using or used the
banks credit card. The data has customers demography information along with a flag stating if the
customer was a fraud customer or not. Using this historical data she wishes to identify if the new
applicant for the credit card would fall under fraud category or not so as to improve the credit card
approval process and add more checks for applicants falling under the fraud category.
The manager approaches the analytics team with the problem and shares the data which has details.
The analytics team gets the data and then explores the data to treat the data for missing values and
outliers. The team comes out with visualization. One of the visualization is shown below –
This chart plots all the variables indicating if the variable has missing values or not in the data.
The analytics team then does statistical analysis to predict the “Fraud” customers based on the
historical data. The team uses different algorithm for Classification to predict the fraud customers
and based on different parameters comes out with the best classification method to be used. Using
this, the manager would know if a new applicant falls under the fraud category so that more
checking can be done on the applicant before taking any decision on the credit card application.