Abstract
In this paper, the task of fraud detection using the methods of data analysis and machine learning based on social and transaction graphs is considered. The algorithms for feature calculation, outlier detection and identifying specific sub-graph patterns are proposed. Software realization of the proposed algorithms is described and the results of experimental study of the algorithms on the sets of real and synthetic data are presented.
Highlights
At present fraud is a major threat that is increasing every year
Manual review remains prevalent among the means of fraud detection
Capgemini [7] claims that fraud detection systems using machine learning and analytics minimize fraud investigation time by 70% and improve detection accuracy by 90%
Summary
At present fraud is a major threat that is increasing every year. The global economic crime survey of 2018 carried out by PricewaterhouseCoopers [1] found that almost half (49%) of the 7,200 companies they surveyed had experienced fraud of some kind. Machine Learning can be used to predict fraud in a large volume of transactions by applying cognitive computing technologies to raw data. Regression analysis tends to become more sophisticated when applied to fraud detection due to the number of variables and size of the data sets. It can provide value by assessing the predictive power of individual variables or combinations of variables as part of a larger fraud strategy. Graph databases allow blocking suspect and bogus accounts before they have taken any fraudulent action Another important trait that makes graph database a valuable addition to any fraud prevention solution is its inherent speed in calculating relationships. Graph database enables quick extraction of new insight from large and complex databases to help uncover unknown interactions and relationships
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have