This paper deals with the data mining techniques used to combat fraud in different industries. The main challenges in fraud analytics have been addressed and the type of fraudsters involved have been studied in the light of different situations. The two main types of data mining methods i.e. supervised and unsupervised learning methods used in fraud analytics have been intensively reviewed with the help of existing literature. This is aimed to help in recognizing which data mining techniques help the most in fraud analysis in respective domains.