A Comparison and Integration of Classification Techniques for the Prediction of Small UK Firms Failure
Chrysovalantis Gaganis, Fotios Pasiouras, Alexander Tzanetoulakos · SSRN Electronic Journal · 2006
In this paper we compare the efficiency of five classification techniques namely, discriminant analysis, logit analysis, Utilites Additives Discriminantes (UTADIS), Multi-Group Hierarchical DIScrimination (MHDIS), and Support Vector Machines (SVMs) in predicting small firms failure. We then investigate the efficiency of integrated models developed through a majority voting rule and stacked generalization. The sample consists of 984 small UK firms, half of which failed between 1997 and 2004. The models are evaluated both in terms of their prediction accuracy, as well as with ROC analysis.