ANALYSIS OF IMPACT OF USING THE TREND VARIABLES ON BANKRUPTCY PREDICTION MODELS PERFORMANCE

The main objective of this paper is to analyse the impact of trend variables on the predictive ability of the models constructed using two methods: discriminant analysis and decision tree technique. The second objective is to develop a new model with prediction accuracy higher by at least 10% in comparison with other models being currently used in the Slovak business environment (Altman model, Index IN05). After analysing and comparing these methods, we came to the conclusion that the most suitable method for developing the model was the decision tree technique. Using this technique we were able to extract classification rules for bankruptcy prediction and achieve predictive ability of about 85% which, in comparison with other models, showed higher predictive performance by about 10%. Moreover, we confirmed that by applying the dynamic approach predictive ability of the decision tree increased; however we did not derive the same result using the discriminant analysis method.