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

從變更項目利益資料庫中探勘高利益項目集之研究

黃永吉; Huang, Yung-ji 李建億; Chien-I Lee; 數位學習科技學系碩士班 2011

線上取得

  • 題名:
    從變更項目利益資料庫中探勘高利益項目集之研究
  • 著者: 黃永吉; Huang, Yung-ji
  • 李建億; Chien-I Lee; 數位學習科技學系碩士班
  • 主題: 變更項目利益值; 資料探勘; 利益探勘; 高利益項目集; Data mining; Utility mining; High utility itemsets; Varying item utility value
  • 描述: 關聯規則探勘是資料探勘領域中的一個重要研究議題,利益探勘(utility mining)延伸傳統的關聯規則,不止考量商品有無購買的二元關係。在現實生活中,使用者所感興趣的項目集是找出對公司最有價值的商品組合,然而頻繁出現的項目集並不一定能滿足這類需求。因此,利益探勘將商品購買數量以及商品獲利(profit)的概念加入探勘工作,找出對整體利益貢獻具影響性的高利益項目集。在過去利益探勘的研究中,商品利益被視為固定不變的,但是在實際應用上,商品的獲利利益會因為許多情況而變更,例如:原物料的上漲、過季商品促銷拍賣…等。本研究目標在於可變動商品項目利益的環境之下,有效探勘高利益項目集,因此,提出VIUV(Varying Item Utility Value)演算法。對於不斷變更項目利益的資料庫,使用一個有效的過濾機制,以排除購買數量少以及獲利利益偏低的交易記錄。由實驗結果可以發現VIUV與尋找HUI(High Utility Itemsets)方法其中的Two-Phase演算法相較之下,在同樣的最小利益門檻值之下,能有效找出高利益項目集。最後,經過實驗多種模擬資料,證實本論文所提出的方法有良好的效能表現。
    Association rule mining is one of the research issues in the field of data mining. Utility mining extends the traditional association rules, which considers are not only the binary purchase relationship. In real-life applications, users are interested in identifying the company''s most valuable product package. However, the most frequent item-set may not the most valuable product package. Therefore, the work of utility mining is to draw purchase quantities and profits into mining concept so we can find the high utility item sets under the whole profit consideration from the transaction records database. In the past research, products benefits are regarded as fixed, but in practical applications, profits are may changed because of many situations such as raising prices on materials, last season’s merchandise, etc. So we propose a varying item utility value high utility itemsets (VIUV) algorithm by applying more effective mechanism to remove transaction record of purchase products with lower-quantity and lower-profit items based on varying item value database and high utility item sets mining.Compared to the HUI (High Utility Itemsets) Two-Phase algorithm, the experimental results can be found VIUV can dig out high utility item sets effectively while setting the same minimum utility threshold. At last, the experimental results on synthetic datasets also show the proposed approach has a good execution time for mining work.
    碩士
  • 建立日期: 2011
  • 格式: 121 bytes
    text/html
  • 語言: 中文
  • 識別號: http://nutnr.lib.nutn.edu.tw/handle/987654321/1799
  • 資源來源: NUTN IR

正在檢索遠程資料庫,請稍等