Customer Profiling: used internal and bureau scores to identify high risk and low revenue pockets

Residual Analysis: used regression and correlation with 1000+ variables

Used variable v/s DV(Dependent variable) curves to find the optimum cut-offs

Successfully implemented for 750K quarterly acquisitions, reducing exposure by $600MM

Conducted Sensitivity Analysis to identify high and low balance/sales sensitive accounts

Calculated guardrails on the Line for  each score band to keep the line within bounds

Carried out the Swap-In /Swap-Out Analysis for existing CLI program to make  it more customer-centric

Built a risk segmentation model to segregate customer behaviour using 1000s of variables

Final strategy was implemented on 24 million card holders with a potential to decrease $700K in losses

Identified the target population by customer profiling

Analysed segmentation variables for the line assignment grid

Predicted  a cumulative 5 year EBIT of $340 per account

Performed quarterly validations of all Risk and Segmentation models to ensure model reliability  for  underwriting

Computed model score on the latest available data and compared against benchmark and development models using KS, population stability, and accuracy

Generated annual validation reports for external audits

Administered MIS for the quarterly validation reports of risk and segmentation models

Built and maintained the model inventory for OCC audits