A Hybrid Approach Using A Back-Propagation Neural Network and Self-Organizing Map for Bankruptcy Type Classification

The prediction of bankruptcy has been steadily studied in the accounting and finance field. In a previous study on bankruptcy, many researchers have focused on developing a bankruptcy prediction model to prevent bankruptcy from occurring. However, there are few studies that classify the specific bankruptcy type. We propose a hybrid approach using a backpropagation neural network (BPN) and selforganizing map (SOM) for the classification of bankruptcy type. We develop a back-propagation model for bankruptcy prediction and construct a Self-organizing map model to divide bankruptcy data into several types. The experimental result shows that each of five bankruptcy types has different characteristics according to eight financial ratios used in this study. By using the proposed approach, it is possible to perform both bankruptcy prediction and bankruptcy type classification.