A Combination of Two Classification Techniques for Businesses Bankruptcy Prediction

Abstract

The scientific community has demonstrated that for the bankruptcy prediction, different techniques have different advantages on different data sets and different feature selection approaches. This subject has attracted a lot of research interests as it is one of the major preoccupation of accounting specialists, business analysts and information systems developers. Because there is no prediction performance via the techniques of artificial neural networks, we have explored this topic, and divided the prediction performance using different techniques into two important parts: 1) bankruptcy prediction and 2) non-bankruptcy prediction. In this short paper, we have built a combination of two well known classification techniques, namely the decision tree and the back propagation neural network. In our opinion, this combination (hybridization technique) provides an approach which inherits advantages and avoids disadvantages of different classification techniques. We described the research results, and demonstrated our expectations.