Going concern prediction using data mining techniques

Abstract

Going concern is a fundamental concept in accounting and auditing and the assessment of a firm's going concern status is not an easy task. Several going concern prediction models based on statistical methods to assist auditors have been suggested in the literature. This study explores and compares the usefulness of neural networks, decision trees and logistic regression in predicting a firm's going concern status. The sample data comprise financial ratios for 165 going concerns and 165 matched non‐going concerns. The classification results indicate the potential usefulness of data mining techniques in a going concern prediction context. Further, the decision tree going concern prediction model outperforms the logistic regression and neural network models. Data mining techniques such as neural networks and decision trees are powerful for analysing complex non‐linear and interaction relationships, and hence can supplement and complement traditional statistical methods in constructing going concern prediction models.