Develop a default prediction scorecard solution to identify the predict the defaulters at the time application for loan.
- Profile data was merged with the transaction data to create a single snapshot of data.
- In Data preparation-data audit reports were created having detail report on Missing value treatment, Outliers detection and treatment.
- Detail trend analysis like correlation with target, Cross tab to see the relationship, binning to investigate inter attribute relationship was done.
- Features were created using statistical methods like multivariate analysis, correlation analysis, binning using decision treeetc. Business knowledge was also used with help of discussion with client in features creation.
- Now business has system to identify the risky customer at the time of loan application itself.
- The default rate of the customer have gone down with more than double.
- Business is able to risk management in a better way by optimizing the use of insurance.