Slicing through a forest of fraud with deep analytics
An Australian telecommunications firm was facing rising customer dissatisfaction and resultant revenue loss due to sales malpractices by some agents at its retail stores and contact centres. This case details how Valerie Coker, Senior Manager for Business Analytics and Fraud Prevention leveraged the company’s 15 year partnership with Infosys BPM to identify and stem the fraud, saving millions of dollars for the firm.
Approach summary:
- Firmographic data collection and transformation
- Development of UML amalgamation model
- Application of Isolation Random Forest algorithm
- Implementing supervised machine learning
- Cross validating model performance with feedback loops
Key benefits:
~100% review coverage of stores and contact centers 
~3.2 Mn AUD identified for loss and remediation 
~400,000 AUD identified for commission payout clawbacks 
