Identifying Consumer Financing Fraud

Identifying Consumer Financing Fraud

Client’s Challenge

➢Unlike other types of fraud such as credit card fraud where anomalous transactions can be identified based on their value and location, consumer financing fraud is more systemic in nature
➢Fraud can be discovered by viewing several transactions together over a period of time
➢At times finding such fraud is humanly impossible as auditor has to go through 100K transactions each month

Analysis & Solution Approach

➢AI systems can be trained to analyze millions of data points in real time, continually scan systems and networks for vulnerabilities, and identify anomalies, unusual patterns and suspicious behaviors.
➢The solution comprised of rules based engine (called the hard filter) followed by machine learning filter
➢The hard filter implemented 11 different rules for already known anomalous conditions. After passing through the hard filter, some transactions are tagged as anomalous and are rejected in the first pass.
➢However, rules based filter was incapable of finding systemic fraud across multiple transactions.
➢Machine Learning filter implemented models capable of identifying systemic fraud.

Benefits Delivered:

✓ Once the models were ready, they were productionized by wrapping them into RESTful APIs. A GUI was developed to run models for monthly transactional data and to show results in report format.0
✓ Millions of transactions have already been processed by delivered solution. The system has been able to detect fraudulent transactions with close to 75% accuracy.
✓ Model accuracy has been increasing with more transactions being processed each month. The solution has led to 80% reduction in cost of auditing.


It’s simple.