A New Artificial Bee Colony Based Clustering Method and Its Application to the Business Failure Prediction

The development of business failure prediction system to prevent the significant loss of social costs caused by the companies´ unexpected bankruptcy is a popular investigation issue. Because of the constraint on the statistic assumptions, the forecasting models established by traditional statistic methods have some limits in its identity. Therefore, in recent years various algorithms imitating of biological behavior have been proposed for improving the accuracy of forecasting models. In this paper, a new artificial bee colony (ABC) based clustering algorithm has been proposed for replacing with previous clustering method to group forecast value by homogeneous. Furthermore, the rough set theory has been utilized to deal with the uncertain data and provide the decision rules and classification results. In the simulation test, the data of listed companies in Taiwan between 1977 to 2011 years have been sampled. Finally, there are 57 companies with bankruptcy crisis have been sifted to verify the effectiveness of the proposed methods. The simulation results indicate that the accuracy of developed business distress early-warning model is much better than that of other methods, especially in the last year of crisis occurrence.