Application of cluster analysis for the assessment of the share of fraud victims among bank card holders

In this paper, we present a method for the assessment of the share of cardholders most prone to various types of bank fraud (i.e. fishing, vishing, skimming). For this purpose, a forecasting information system has been designed. It is based on a clustering module used for output of a certain set of cluster indices that depend on the percentage of aggrieved clients in the training sample. The k-means method is used for clustering. The initial coordinates of centroids are defined using advanced k-means++ algorithm.