James Taylor's Decision Management

Insurance and Technology had a great set of posts on fraud detection and analytics - Emerging Technologies Fight Claims Fraud. In particular I liked the fact that they covered both the use of data mining and analytics to support investigation and the use of predictive analytics in individual transactions. While I tend to focus on the use of analytics in the context of transactions, there are many ways to use analytics in fraud detection:

• Helping a human investigator understand what the data they have is trying to tell them
• Helping a business owner understand cut-offs and thresholds that should be embedded in their business rules to route potentially fraudulent claims to investigators
• Showing the impact of changes to the way claims are handled to avoid unexpected outcomes - like too many referrals for review
• Calculating the probability that a specific transaction is fraudulent so that it can be dealt with appropriately