(Received 6 December 2011; accepted 24 April 2012)
Published Online: 2012
| ||Format||Pages||Price|| |
|PDF (184K)||8||$25||  ADD TO CART|
Traditional statistic models for financial distress are subject to constraints which may lead to imprecise prediction. To contribute to the issue, we construct a two-staged integral model by applying a stepwise regression analysis and a data-mining approach. Specifically, we employ stepwise regression and rough set analysis in feature selection to sieve out variables, and perform decision tree, neural network, and logistic regression analysis to classify firms with financial distress. The findings show that the rates of accuracy for the combinations in descending order are stepwise regression-logistic, stepwise regression-neutral network, stepwise regression-decision tree, rough set theory-neutral network, rough set theory-decision tree, and rough set theory-logistic.
Dept. of Accounting, Chinese Culture Univ., Taipei,
Dept. of Accounting, Shih Chien Univ., Taipei,
Stock #: JTE104584