Visualisation of Complex Business Data: A Neural Network Approach

Reliable and well-audited financial statements attract the capital that finances business. Analytical auditing plays an important role in assisting the auditor in determining the nature, timing and extent of his or her substantive testing and in forming an overall opinion as to the reasonableness of recorded account values. It is used to improve the efficiency of auditing. Basically, in an analytical auditing one compares expected relationships among data items to actual observed relationships. This paper shows how neural networks, especially Kohonen's self-organising map (SOM) can be used in analytical auditing when auditing monthly account values. The SOM is used for clustering monthly data sets. Neural network systems are based on computational intelligence. The purpose is to show how the data sets of various accounts and various years form their own groups. We found that the SOM can add value to auditors in the analytical process: it is a tool for classifying and clustering data sets that reveals if some cluster contains data that a priory should not be in it. Therefore, it can be used for signalling unexpected fluctuations in data. Furthermore, the SOM is a possible technique embedded in the continuous auditing tool.