長久以來許多研究專注在金融市場的關聯性,嘗試在金融市場裡找出可靠依據的原則。根據這些原則來提供相關的訊息或是預測結果來供投資者或是決策者參考,並且在金融市場中挖掘一些不容易發現且含有意義的資訊。本研究利用資料探勘中關聯法則的技術以及灰關聯分析,來探討台灣股市交易市場中有相互影響的公司是否有連動關聯性,以提供投資人作為判斷投資的一個依據。 本文針對台灣電子類股之3百多家上櫃上市公司,並依人力資源網站所提供之公司分類排名,找出欲探究標的公司來分析在區段交易日內之相互關聯性,並搭配灰關聯分析以驗證所探勘出來之規則是否具可信度。研究中將標的個股每日收盤價轉換成漲跌幅,在不同支持度與信賴度情況進行分析,實驗結果可具體發現不同資料庫其探勘出之關聯法則之多寡。此外,選擇不同興趣量測值,更可從關聯法則篩選出具意義的規則,此等有意義的規則搭配視覺化分析,可進一步確認不同公司間其股價變化之關聯性。最後,本論文利用灰關聯分析技術來計算具有意義的規則間之關聯性,結果顯示關聯度高低確實吻合資料探勘之結論。
Many researchers have focused on finding the correlation of stock markets to provide meaningful rules for the financial market. The rules can be used as the references or predictions to investors or decision-makers to discover some meaningful information which tends to be overlooked. This study utilizes the data mining technique and grey relational analysis to find correlations among companies listed in the Taiwan Stock Exchange Market. Among three hundred plus listed companies in Taiwan’s electronic industry, based on the categories or rankings from the websites hosted by human resource agencies some target companies were chosen for our analaysis. The stock indices were transformed into percentage to find the correlations. Various support and confidence thresholds are applied to discover the association rules. Interesting measure is also used to further filter the uninteresting rules. The daily stock indices of highly related companies are plotted for visualization comparison. Grey relational anaysis is used to calculate the relational degrees among the stock indices. The rank of grey relational degrees conforms to the finding in data mining.