skip to main content
資源種類 顯示結果: 顯示結果: 查詢種類 索引

利用三個門檻值以有效探勘遷移樣式之研究

李文欽; Lee, Wen-chin 李建億; Chien-i Lee; 數位學習科技學系碩士班 2008

線上取得

  • 題名:
    利用三個門檻值以有效探勘遷移樣式之研究
  • 著者: 李文欽; Lee, Wen-chin
  • 李建億; Chien-i Lee; 數位學習科技學系碩士班
  • 主題: 資料探勘; 關聯法則; 準頻繁項目集合; 樣式遷移; data mining; migration pattern; pre-large; association rule
  • 描述: 由於電腦運算能力的快速精進與資料庫系統的大量使用,如何讓使用者從龐雜的資料中取出隱含其中的知識已經成為許多專家學者所專注的研究議題。專家學者過去通常假設資料庫是靜態的、是不會加入新交易的。因此,許多已提出的探勘演算法也都假設資料庫是靜態的。但在真實環境的動態資料庫中常會加入新的交易記錄,在此環境下,原先常見的交易樣式(如:顧客的常見購買行為)就會變動,是為樣式遷移(Patterns Migration)。也就是說,在交易資料庫中,項目集合因為新交易的加入會由頻繁變成不頻繁或是由原來不頻繁變為頻繁的可能。對資料探勘的使用者來說,感興趣的可能正是這些發生遷移的項目集合。傳統資料探勘演算法在探勘樣式遷移時須先探勘更新後的資料庫才能再對樣式做比對。當更新交易筆數不多的狀況下探勘更新資料庫所花的時間幾乎與探勘原始資料庫相同;如此探勘樣式遷移將是沒有效率的。 因此,本論文提出利用三個門檻以有效找到樣式遷移的演算法稱之為FM演算法。當有新增交易加入時,PM演算法可以藉由以探勘過的歷程或是資訊,可以找出發生樣式遷移的項目集合而不需完整探勘更新後的資料庫。實驗結果顯示PM演算法比起傳統演算法能更快且更有效的找出發生遷移的項目集合。
    Due to the increasing use of very large databases and data warehouses, mining useful information and helpful knowledge from transactions is evolving into an important research area. In the past, researchers usually assumed databases were static to simplify data mining problems.But, in real-world, database were not always static and would update containtly. Hence, the pattern might by change and could be treat as patten migration. Some frequent patterns might be infrequent or some infrequent pattens might be frequent after updating newly transactions. However, users might be very interested in this patterns migration. The conventional batch-mining algoritm have to re-preocess the entire updated database to find and to compare the different frequent patterns. If the newly database is small, the computational time is almost the sme with original database. That is inefficient. The proposed algorithm, FM algorithm, can dicover migration patterns by applying three thresholds. FM keeps the pattern information of previous mining process to efficiently find migration patterns without re-scanning the entire database. The experiment shows that FM algorithm was more efficient than Apriori to find migration patterns.
    碩士
  • 建立日期: 2008
  • 格式: 121 bytes
    text/html
  • 語言: 中文
  • 識別號: http://nutnr.lib.nutn.edu.tw/handle/987654321/3414
  • 資源來源: NUTN IR

正在檢索遠程資料庫,請稍等