MINING THE TELECOMMUNICATION DATA FOR FRAUD DETECTION ON CALL DETAIL RECORD
No Thumbnail Available
Date
2019
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
UMT, Lahore
Abstract
The telecommunication industry is growing rapidly and telecom companies are collecting and recording huge amounts of data that can be used to extract interesting patterns. Different data mining methods can be applied and based on the knowledge obtained, we can set up new strategies for tasks like fraud prevention, promotions, offering bundle packages for prepaid and post-paid customers, and many others. Telecom companies in Pakistan provide a variety of services to win over their subscribers because of the tough market competition. Fraud is a multi-billion-dollar problem around the globe. Telecommunication fraud causes a huge loss of income and it can affect the performance and credibility of telecommunication companies. The most difficult problem that the industry faces is that fraud is dynamic. Over the years, fraud has become a major issue to the extent that losses to telephone companies are measured in terms of millions and billions of American dollars. Fraud adversely impacts the telephone company in four ways shareholder perceptions, customer relations, financially and marketing. In this work, our main focus is on analysis of outgoing calls data to identify potential fraud cases. We further look at the vulnerability of the system and predict if the same type of fraud can occur in future. Our proposed approaches are based on data mining and machine learning techniques including SVM, Artificial Neural Networks and Naïve Bayes. We achieve an accuracy of 93.19% for vulnerability prediction and for fraud detection we achieve an accuracy of 93.2%.