Forge Detection In Credit Application
dentity theft is a crime whereby the criminals impersonate individuals. Credit application fraud is a prominent
example of identity crime. Nowadays, credit application fraud is widespread and has reached a mass level of fraudsters.
Even the fraudsters are highly organized, sophisticated and their visibility patterns are quite distinct and tend to change from
each other. The existing system of business rules, score cards and known fraud matching are based on the non-data mining
approaches and supervised algorithms have certain limitations. To develop a data mining based approach to encounter credit
application fraud, a multilayered detection system is proposed. It uses additional layers: Communal detection (CD) and spike
detection (SD). CD devises a way to find the real social relationships in a dataset and works upon fixed set of attributes. SD
determines spikes in duplicates and acts on variable set of attributes. A sudden and sharp rise in spikes increase the suspicion
score. Together, the combination of CD-SD addresses the issues of scalability, imbalanced class and changing behavior.
Index Terms- Identity theft, Data mining based approach, adaptive, Imbalanced class, Communal detection, Spikes, Spike