本論文基於約略集合,K-均值分群法和灰色關聯分析理論,並結合技術分析為投資者提供一套自動篩選股票系統。首先,我們收集上市櫃公司報表中財務比例,並選擇約略集合的條件屬性與決策屬性,然後再利用約略集合的重要度選擇重要的條件屬性。接著,進行基金的權重分配,使用灰色關聯度的方法來獲得更好的投資回報率。最後,通過技術分析找到合適的購買和出售時機。本研究針對台灣新經濟資料庫股票上市公司財務報表數據,篩選出最佳的股票投資組合。在台灣的實證結果:四年期間(2006-2011年),平均每年的報酬率為21.43%。
This thesis offer investors an automatic stock portfolio selection system with technical analysis based on a Rough set, K-means and Grey Relational Analysis (GRA). First, we collect relative financial ratio datum as the conditional attributes selection and then use significant value of rough set for choosing the more important conditional attributes, and rough set for figuring the best portfolio out. Then, conduct fund weight distribution using the grey relational grade method to get the better return rate for investment. Finally, technical analysis was adopted to find suitable buy and sale timing. This study focuses on listed companies of Taiwan stock to sift the optimal stock portfolio out applying the financial statement datum from the New Taiwan Economy database(TEJ). The empirical result in Taiwan: During four years (2006-2011), the average annual rate of return was 21.43%.