Extracting Rules of Initial Returns Using Attribute Selection and Entropy-Based Rough Sets in Electronic Firm

This paper briefly forecasts initial returns in initial public offerings (IPOs) market of Taiwan stock trading systems by attribute selection and entropy-based rough set. It is very important for investors that correctly predict initial returns from trading systems because of the more accurate prediction, the more gain profit. In this paper, we use attribute-selecting based method to enhance accuracy of classifier, and use entropy-based method to discretize attributes for enhancing rough set classifier. The practical IPOs dataset is employed in this case study to illustrate the proposed approach. From the results, the proposed approach has three advantages: (1) improves accuracy, (2) reduces attributes and (3) generates fewer rules. Furthermore, the performance is superior to the listing methods.