Using Financial Ratios with Artificial Neural Networks for Bankruptcy Prediction

Bankruptcy prediction has become an emerging research area in recent years in the field of business financial management. This study seeks to improve the application of artificial neural networks using a novel approach to feature engineering. Two of the main factors in a successful prediction model are the model's accuracy in terms of identifying companies that will go into bankruptcy and how early those predictions can be made. Business financial ratios are used as features in computational bankruptsy prediction. In this study data from annual reports of over 14,000 UK and Irish companies for a period of 10 years were used to derive 29 financial ratios, that have not been used together in bankruptcy prediction, in previous work. These comprehensive set of financial ratios provided predictors for the novel deep neural network models we created. The results reveal the possibility of significantly better and much earlier prediction of bankruptsy with the proposed model compared to the approaches proposed in previous studies.