The prediction of financial distress using structured financial data from the internet

This paper uses Þnancial structure data via internet network to enable the prediction of Þnancial distress. Financial distress is deÞned as a Þrm that has entered into liquidation, receivership or is declared as of negligible value by the Inland Revenue. A sample of distressed and nondistressed quoted companies listed on the London Stock Exchange for the last 20 years (1980 1999) was obtained for this empirical study. Initially we use four single classiÞers discriminant analysis, logistic regression, neural networks and decision tree C5.0 each based on three feature selection methods for predicting Þnancial distress. Of the three feature selection methods human judgement, ANOVA and factor analysiswe found that ANOVA method performs better than the human judgement method and factor analysis in all of the classiÞers except discriminant analysis. Among the individual classiÞers, decision trees and neural networks were found to provide the better total accuracy. Finally, a cost sensitive hybrid method is developed to increase the prediction performance.