The Methodology of Risk Decreasing and Fraud Operations Avoiding with Payment Cards of International Payment Systems - Essay Example

The methodology of risk decreasing and fraud operations avoiding with payment cards of International Payment Systems (IPS). In fact, the payment (credit) cards became more and more widespread all over the world. Therefore the ratio of fraud operations with them becomes higher. That’s why the developing of effective methodology of risk management looks like to be “a must”. Frankly speaking, you will not be able to find enough necessary information anywhere, because the methods of fraud detection are security-guarded and frequently changed.

Here we tried to observe briefly the most fundamental methods of statistical analysis, systems analysis of multiple risks and Bayes confidence net theory theory. The main idea of statistical analysis is creating a universal antifraud filter: mechanism, which can detect fraud operations, can be used on all levels from the merchant, then payment gateway, and, at last, bank-acquirer. Content analysis meets these requirements. Content analysis is a quantitative and qualitative method of documents investigating, characterized with conclusions objectivity, procedure rigidity and means quantification processing with further results interpretation.

For fraud operations detecting and negative results avoiding we can use some methods of content analysis: variance analysis for some independent factors influence on feature under review; cluster analysis for object and their features classification; longlinear analysis for statistical testing of hypothesis about the system of simultaneous even and plural relations in features group; analysis of causes for causal relations modeling among features with the help of statistical equation systems; regression analysis for regression dependence investigation among dependent and independent features; factor analysis for getting general information about the relation structure among the features of object under consideration with concealed factors detaching; correlation analysis for dependence detecting among numerical variates, one of which also depends on many chance factors.

The great quantity of these methods gives us an opportunity to choose the most suitable for each specific task, for each kind of database. Systems analysis of multiple risks is used when we have no considerable database of certainly fraud operations, which represents different fraud kinds. The idea of using this methodology lied in developing of system classification by degree and level of risk and situation recognition of given classification. Alaric, a company, which specializes on banking systems, insists that fractal systems of fraud prevention, based on adjusted rules with using Bayes logic, are more effective then neuronets.

But the high result production of there systems can be reached only after 4 months for banks with great quantity of emitted cards and high branched acquirer system. The neuronet can help to create a mechanism of operation stream evaluation, which will be based on special rules. Firstly, these rules are defined by experts, later changed and corrected with system. This function is necessary system component, as the fraud methods are changed as times goes by, and detecting them mechanically, we’ll miss more refined and, surely, more detrimental for a bank.

The process of Bayes confidence net theory is learned in 2 phases: Identification of net topology. Especially it’s concerned missing chains and movement direction. Learning of analogous values for given net topology. The second task is trivial enough. For topology identification can be used next algorithms of local optimization with calculation of partial derivatives of first order; local optimization with calculation of partial derivatives of first and second orders, for example, Newton method; stochastic optimization algorithms, for example, Monte-Carlo method, algorithms of global optimization etc.

References 1. P Chan, W Fan, A Prodromidis & S Stolfo. 1999. Distributed data mining in credit card fraud detection, IEEE Intelligent Systems. 14(6): 67–74. 2. J. Stolfo, David W. Fan, Wenke Lee and Andreas L. Prodromidis.1999. Credit Card Fraud Detection Using Meta-Learning: Issues and Initial results, IEEE Intelligent Systems, 20(7): 67–85.