A Comparison among Artificial Neural Network, Discriminant Analysis and Logestic Regression Techniques for Bankruptcy: A Case Study of Kerman's Firms

Makyian, Nezamoddin; Almodaresi, Mohammadtaghi; Karimi Takloo, Salim

A Comparison among Artificial Neural Network, Discriminant Analysis and Logestic Regression Techniques for Bankruptcy: A Case Study of Kerman's Firms

Authors

nezamoddin makyian ¹

mohammadtaghi almodaresi ²

salim karimi takloo ³

¹ Yazd University, Economics Faculty

² yazd university, electrical engineering faculty

³ kerman university

Abstract

One of the main issues in financial management is choosing the best way of utilizing investment. Investors would like to invest their capitals in a way to minimize their risks. Bankruptcy is one of the risk factors which affect the decision of investors. Prediction of bankruptcy can help investors to reduce the risks in the capital markets and recognize the best opportunities for alternative investment.
This study aims to predict the bankruptcy of companies by using the technique of Artificial Neural Network (ANN). Moreover, discriminant Analysis and logestic regression techniques are employed to compare the results. The data used in this study covers the firms in the Kerman Province of Iran over the period 1975- 2007.
The results show that ANN model perfom much better than the discriminant analysis and logestic regression techniques. Moreover, the results confirm that the accuracy of ANN model is higher than the discriminant analysis and logestic regression techniques for predicting of bankruptcy. The analysis also shows that none of the firms will bankrupt in the year after the period covered in this study.

Keywords

Bankruptcy Prediction

artificial neural network

Logestic Regression

discriminant analysis