Machine Learning
Context-Based Text Mining for Insights in Long Documents
PAKM '08 Proceedings of the 7th International Conference on Practical Aspects of Knowledge Management
Hi-index | 0.00 |
We presents some empirical results of a study regarding financial ratios as predictors of Japanese corporate bankruptcy based on a large sample of bankrupt and non-bankrupt firms' financial data. In this study, variable as predictors of bankruptcy had been selected by three Al-based data mining techniques and two conventional statistical methods, Logit analysis and Stepwise. After the selection of a set of variables for every method, discriminant power of each set was compared to verify the most suitable data mining technique to select financial variables. Finally, the study concludes that a set of variables selected by Logit analysis (with logit model) indicated the best discriminant power, more than 87% accuracy.