CASPIAN JOURNAL
MANAGEMENT AND HIGH TECHNOLOGIES
Development of a proactive method for fraud detectionin in Telecommunication Company
 |
Read | Nguyen Tuan Anh, Shcherbakov M.V., Tran Van Phu, Kravets A.G. Development of a proactive method for fraud detectionin in Telecommunication Company // Caspian journal : management and high technologies. — 2016. — №4. — pp. 43-52. |
Nguyen Tuan Anh - post-graduate student, Volgograd State Technical University, 28 Lenin Av., Volgograd, 400005, Russian Federation, anhtuank37@gmail.com
Shcherbakov M.V. - D.Sc. (Engineering), Professor, Volgograd State Technical University, 28 Lenin Av., Volgograd, 400005, Russian Federation, maxim.shcherbakov@vstu.ru
Tran Van Phu - post-graduate student, Volgograd State Technical University, 28 Lenin Av., Volgograd, 400005, Russian Federation, vanphu.vstu.russia@gmail.com
Kravets A.G. - D.Sc. (Engineering), Professor, Volgograd State Technical University, 28 Lenin Av., Volgograd, 400005, Russian Federation, agk@gde.ru
The paper presents the results of research on consumer fraud detection in telecom. The basic idea is in proactive detection of fraud situations based on users’ behaviour analysis. The approach based on the assembling of machine learning algorithms is proposed. Synthetic data sets containing Call Detail Records are used for training and testing algorithms. The choice of the best algorithm is made based on searching of hyperparameters with cross-validation. It is shown, that SVM is the most efficient algorithm for the certain data set. The distributed architecture of the fraud detection system was presented which implements the described approach.
Key words: телекоммуникационное мошенничество потребителей, обнаружение мошенничества, машинное обучение, обработка данных в реальном времени, пакетная обработка данных, методы интеллектуального анализа данных, методы принятия решений, telecommunication fraud, fraud