In this chapter, we will zoom out on fraud analytics and discuss some issues from a much broader perspective. The availability of more and more data across a diversity of channels along which a customer interacts with a firm necessitates a thorough reflection about two key issues: data quality and privacy. This is especially relevant for a mission critical application like fraud detection. A key usage of analytical fraud-detection models is the calculation of both expected and unexpected fraud losses, which serve to determine a company's provisions and equity buffers. A thorough economical insight into the total cost of ownership and return on investment of analytical fraud models is also required from both a managerial and investment perspective. Both will be key inputs to decide whether a company should build the analytical skillset in house, or consider outsourcing as an alternative. We also briefly zoom into some modeling extensions such as text analytics and forecasting, and discuss the impact of the Internet of Things on fraud. The chapter is concluded with a discussion about corporate fraud governance.
A broad perspective toward data quality covers both the fitness of data for its intended use as well as the correctness of the data with respect to representing, describing, or measuring real-world entities or events. The fitness of its use will not be discussed in depth in ...