透過您的圖書館登入
IP:3.21.104.109
  • 期刊

以多重彈性倒傳遞類神經模型探討台灣4G概念股股價的變動

Applying Multiple Resilient Back-Propagation Neural Model to Explore 4G Stocks Market

摘要


股票為台灣投資人常使用的投資工具,股價反映公司在市場的價值,股價易受到新產品的推出而產生波動,近年來通訊技術已發展到4G 技術,我們試著從歷史資料中尋找此新技術對於該產業股票價格的影響,本研究利用人工智慧的資料探勘技術來選出與股價有關連的指標,並建立一個預測的模型,提供投資決策參考。本研究以4G 概念股中三檔股票為例,以分類與迴歸樹篩選與收盤價有相關之技術指標,使用模糊分群法將資料分群,再使用彈性倒傳遞類神經網路探討其變動,最後將此模型與時間計量模型結果進行比較。在實驗中我們發現,各樣本公司篩選出來的技術指標都不盡相同,而倒傳遞神經網路的準確率都優於時間計量模型的結果。

並列摘要


Stock is a common investment tool in Taiwan. The prices of stocks reflect the values of companies. Recently, 4G technology industry has affected he stock prices. In this study, we try to explore stocks market dynamism of three companies. We use classificaion and regression tree to select the influential technical indicators. Then we use fuzzy C-means algorithm to cluster training data. We use Resilient Back- Propagation Neural Network to build predictive models for each cluster of training data to explore the precision rates. In the experiment, we found that the influential technical indicators for every company are various. Also, the proposed approach outperforms the benchmark method.

延伸閱讀