Piotr Chrzan, Ph.D., The Karol Adamiecki University of Economics in Katowice, u. Gen. Pulawskiego 25, Katowice, 40-276, Poland
Company bankruptcy is usually preceded by slow and multi-stage worsening of its financial situation. Therefore, it’s possible to construct prognosticating models aimed at bankruptcy risk evaluation.
The research is aimed at constricting bankruptcy prognosticating models using the same set of data but different classification methods and at comparing those models according to their prognostic futures.
Analyzed data includes financial reports of 50 companies with good financial standing and 50 companies which went bankrupt. All of them operated at Silesian region. Using the same data sets bankruptcy prediction models are constructed based on at least four classification methods: multiple discriminant analysis, neural network, logistic regression and mathematical programming.
Final comparison analysis of the results entitles to conclusion formulation on the relation between theoretical basics of analyzed classification methods and their prognosticating characteristics.