資訊管理學報
邱登裕;徐廣銘;
頁: 73-96
日期: 2008/01
摘要: 方法是結合遺傳演算法與法則式類神經網路，克服利用類神經網路進行財務預測時，其缺乏解釋能力及無法在類神經網路模組中加入專家知識的兩大缺點，而提出一個具有解釋能力的決策模式。期望能透過決策模式的解釋能力讓預測結果能更加取信投資人，並協助投資人進行股市投資，期望能對個股買賣時機提出參考與建議。 本研究針對台積電、聯電與鴻海三支個股來進行買賣時點的探討，並與買入持有法及倒傳遞網路法進行比較。實驗結果顯示，本研究提出的決策模式，其投資報酬率高於倒傳遞類神經網路法及買入持有法。
關鍵字: 決策模式;遺傳演算法;法則式類神經網路;倒傳遞類神經網路;
Abstract: Artificial neural network has been widely applied to predict financial market during the past decades. However, two major defects have limited the development of artificial neural network, the lack of explanation of causal relationship and the scarcity of the integration of expert knowledge. In this study, evolutionary genetic algorithm and rule-based neural network are combined to provide a decision model with explanation. Through the explanation, investors can understand the causal relationship of the prediction result. It can be used to recommend the proper time to buy or sell stocks. An example based on the Taiwan stock market is utilized to evaluate the profit of the proposed decision model. We also compare its performance with those of buy-and-hold method and back-propagation neural network. The experimental results show that the proposed decision model outperforms the other methods.
Keywords: Decision model;genetic algorithm;rule-based neural network;back-propagation neural network;
瀏覽次數: 12320 下載次數: 176
引用 導入Endnote
頁: 73-96
日期: 2008/01
摘要: 方法是結合遺傳演算法與法則式類神經網路，克服利用類神經網路進行財務預測時，其缺乏解釋能力及無法在類神經網路模組中加入專家知識的兩大缺點，而提出一個具有解釋能力的決策模式。期望能透過決策模式的解釋能力讓預測結果能更加取信投資人，並協助投資人進行股市投資，期望能對個股買賣時機提出參考與建議。 本研究針對台積電、聯電與鴻海三支個股來進行買賣時點的探討，並與買入持有法及倒傳遞網路法進行比較。實驗結果顯示，本研究提出的決策模式，其投資報酬率高於倒傳遞類神經網路法及買入持有法。
關鍵字: 決策模式;遺傳演算法;法則式類神經網路;倒傳遞類神經網路;
Investment Decision Model Construction and Exploring the Time for Trading Stocks
Abstract: Artificial neural network has been widely applied to predict financial market during the past decades. However, two major defects have limited the development of artificial neural network, the lack of explanation of causal relationship and the scarcity of the integration of expert knowledge. In this study, evolutionary genetic algorithm and rule-based neural network are combined to provide a decision model with explanation. Through the explanation, investors can understand the causal relationship of the prediction result. It can be used to recommend the proper time to buy or sell stocks. An example based on the Taiwan stock market is utilized to evaluate the profit of the proposed decision model. We also compare its performance with those of buy-and-hold method and back-propagation neural network. The experimental results show that the proposed decision model outperforms the other methods.
Keywords: Decision model;genetic algorithm;rule-based neural network;back-propagation neural network;
瀏覽次數: 12320 下載次數: 176
引用 導入Endnote