Data mining technology is applied to fraud detection to ascertain the scam detection model, to depict the process of creating the scam detection model, and then to begin the data model with any classifier. As e-commerce transactions persist to develop, the allied online hoax remains an eye-catching resource of income for fraudsters. This counterfeit activity inflicts a significant financial hammering to merchants, making online fraud detection a prerequisite. The issue of scam detection is concerned with not only confining the fraudulent activities, but also detaining them as rapidly as possible. This relevance is decisive to shrink financial losses. Cyber crimes are a communal pest and rate our society greatly in numerous ways. The investigation of cyber crime cases has very significant role in police enforcement system in any country. This paper presents a comprehensive study on data mining techniques and its responsibility on detection of cyber crimes in real time applications. Data mining robotically sieves through massive quantity of data to uncover known/unknown patterns that fetches out valuable, innovative perceptions and formulate predictions. Data mining which is alienated into two learning skills viz., supervised and unsupervised is engaged to detect fraudulent asserts. Basically these techniques are used for fraud detection in many sectors such as health , insurance, E-commerce and it goes on.