隨著世界各國經濟的快速發展,經濟現象的預測已經是政經與企業擬定未來政策的重大考量因素,其中又以股價指數預測之相關研究較為重要,包含技術分析、基本分析、產業分析、政策分析和心理分析等等,都是要試圖找出可依循的市場規則,進而獲取更高的報酬。由於目前分類法技術與應用相當成熟,並廣泛應用在萃取資料規則及分類中,因此,本研究運用分類法中決策樹工具ID3搭配模糊理論作為本研究之研究工具,並期望藉由此一學習性的工具,準確預測出股票的買賣點,提供買家較為有利的資訊平台。本研究之研究方法共分為三階段,第一階段為資料前處理的部份,包括逐步迴歸因子篩選與K-means集群分析法,第二階段為模型建立,包括資料糢糊化以及決策樹建立以及案例的測試;由於ID3決策樹主要處理離散型資料,因此本研究藉由模糊化將連續型資料轉換為離散型資料,第三階段則是利用基因演算法進行參數最佳化。經過實驗結果證實,當分群數增加時,股價預測之準確率也相對提升;此外,改變模糊區間確實對於股價趨勢之預測有顯著的影響,當模糊區間越多、越能敏銳的掌握股價趨勢。
Predicting stock data with traditional time series analysis has been proven to be difficult. The time series forecasting in stock market is characterized by data intensity, noise, non-stationary, unstructured nature, high degree of uncertainty, and hidden relationships. Its behavior is more like a random walk. Stock price prediction has always been a subject of interest for most investors and professional analysts. Nevertheless, finding out the best time to buy or to sell has remained very difficult because there are too many factors that may influence stock prices. This paper establishes an evolving fuzzy decision tree model to predict financial time series data in Taiwan Stock Market. This forecasting model integrates data clustering technique, Fuzzy Decision Tree (FDT), and Genetic Algorithms (GAs) to be a trading signals’ decision making system. Financial time series data will be divided into k sub-clusters by adopting K-means algorithm. The GAs is utilized to Evolving the fuzzy term number in FDT so as to improve the forecasting accuracy. Each sub-cluster will generate a different forecasting model, in other words, the fuzzy term numbers evolved by GAs are different. This forecasting model (GAFDT) can help investors to make a better decision for trading stocks.