Identifying intended and unintended errors in financial transactions: a case study

Abstract: An internal audit group of a bank aims to identify the intended and unintended errors in the datasets from various sections such as the stock market. The identification is provided using various heuristics and data analysis techniques. One of thechallenges in the audit group is to find efficient and mostly automated detection techniques. The best identification method would reduce the man-hour needed to process the data and eradicate the errors in detection. In this article, we produce an efficient error detection method based on data mining techniques that alleviates theses difficulties to some extent. We provide a Matlab script that employs the built-in implementation of the hierarchical clustering algorithm and clusters the transaction data from SwedBank. We confirm theeffectiveness of the algorithm and the meaningfulness of the results in financial terms.