2 Fraud detection overview
This chapter covers
- Different ways of classifying fraud.
- Exploring some of the different types of online fraud
- How do we evaluate a fraud detection system?
In chapter 1, we have discussed fraud, fraud-detection, and different ways of designing fraud-detection systems using fixed rules, machine learning, and deep learning. Before we go hands-on into designing some of these fraud-detection systems in this book, firstly we broadly review three different ways of classifying fraud. Using these, we dissect and explore a few major online/internet frauds in this chapter - phishing, identity fraud, online transactions fraud, digital document forgery and social network fraud. This will help us understand the different ways in which fraud attacks can occur. And such an overview prepares us to classify fraud into appropriate categories and accordingly build and use relevant fraud-detection solutions.
Secondly, we walk through the different standard ways of evaluating fraud-detection systems. These evaluation methodologies are universal for all types of fraud-detection systems, be they rules-based or powered by machine learning (or deep learning). A good understanding of an evaluation framework is important to tell apart a good fraud-detection system from a not-so-good fraud-detection system.
Let’s go!