多重資料串流環境序列樣式探勘之應用－以台灣股市為例

Application of Multiple Data Streams Sequential Pattern Mining on Taiwan Stock Market

趙景明(Ching-Ming Chao) ； 楊慧雯(Huei-Wen Yang)

《資訊管理展望》 12卷2期 (2010/12) Pp. 113-132

https://doi.org/10.29715/JIMCSA.201012.0001

摘要

資料變化快速、即時需求提高，資料串流因而興起，其又可分為單資料串流與多重資料串流，多重資料串流可在單一時間處理一整個資料集(Itemset)，提供更即時的分析。因股票資料具有網路公開、數量龐大、更新快速等特點，無疑是具有代表性與實用性的應用，故本研究要建置一個股票探勘系統，運用多重資料串流技術可處理大量資料、即時動態產出分析結果的特色於系統中。將每一檔股票視為一個資料串流，在不同股票間進行多重資料串流序列樣式探勘。從股價的歷史記錄預測未來走勢，週期性探勘不同股票間漲跌的相對順序，幫助投資人掌握即時股市行情，增加獲利機會。

關鍵字

序列樣式探勘 ； 資料串流探勘 ； 多重資料串流 ； 股市分析

並列摘要

Fast changing data and increasing real-time demands have led to the emergence of data streams, which can be categorized into two types: single data stream and multiple data streams. Multiple data streams can deal with a whole itemset at a time to provide more immediate analysis. Since stock data is featured by its open public access on the internet, plenty quantity, fast update and so on. There is no doubt that it is the representative and practical application. This research aims at building up a stock mining system. The characteristic that multiple data streams technology can process large amounts of data and dynamically produce real-time analysis results is applied to this system. Each stock will be regarded as a data stream. We will mine sequential patterns in multiple data streams among different stocks. In this case, the future trend of stock prices can be predicted based on the historical records and the fluctuation between different stocks can be found by cyclical mining, which will help the investors obtain the latest stock market news and then increase profit opportunities.

並列關鍵字

Sequential Pattern Mining ； Data Stream Mining ； Multiple Data Streams ； Stock Market Analysis