Artificial Neural Networks in Predicting a Dichotomous Level of Financial Distress for Uneven Training and Testing Samples

Abstract

To adequately perform their duties, bank lending officers, financial analysts, and auditors must accurately assess companies’ financial performance. Stockholders also have a financial incentive to monitor companies’ financial performance. Accurately predicting financial distress is an important part of the assessment and monitoring process. For decades, these individuals used traditional statistical techniques such as regression analysis, logit regression (LR) models, or discriminant analysis to try to predict which companies are likely to be healthy and which ones will go bankrupt. In the last several years, one can observe a growing interest in the use of relatively new data mining tools such as artificial neural networks (NNs) for the tasks of prediction, classification, and clustering.