A Study on the Optimal Diversification of the Listed Companies in Taiwan─ The Application of Data Mining Technology

• Main Authors: Chia-Liang Yen, 顏嘉良
• Other Authors: Mei -jui Sun
• Format: Others
• Language: en_US
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/35430768356679309890

碩士 === 東吳大學 === 國際貿易學系 === 93 === Abstract This study focus on the optimal diversification of companies listed in Taiwan between 1986 and 2003. We use the research methodology different from before to probe for the related issues about company implement optimal diversification. Here we adopt data mining technology and use it to analyze the related issues about the optimal diversification. The way is to be expected that the company can seek out the proper diversification and can adjust the diversified operating itself. More importantly, it can improve the efficiency of operating and get the better performance to reach the optimal diversification. According to the empirical results mentioned in previous chapter, we can conclude the below points: 1. We use artificial neural networks and decision trees to summarize out those important components and key factors. 2. Companies have to decide what level of the proper diversification does it apply to according to various factors when desiring to know what is the proper level of diversification for an enterprise. 3. When the actual level of diversification value is closed to the predicted value, Sales Per Employee、Net Income Per Employee、Net Sales、Net Income、EBIT and EBITD is significantly increase. However, when the actual value is closed to the predicted value, ROA、ROE and EPS is also significantly decrease. Besides, the explanation ability of regression is not better and it means that R square is very low. 4. We could know that it not absolute to make a well crediting rating or have a little frequency on financial crisis and full-delivery event based on the high level of diversification.