卷期 : 46 / 2
出版年 : 2018/04/01
傳統的資料探勘技術著重於描述性的分析,分析後的資訊用於解釋或說明,近年來的資料探勘整合了多方面的技術,包含了人工智慧、演算法、統計學、資料庫,且因數據量爆炸性的成長而有了新的代名詞大數據。本季刊研究採用已建置完成之軍品料號資料,運用資料探勘技術挖掘有知識,並以相似物件預測欄位可能的輸入值,在使用者輸入前供其參考,減低人為疏失及錯誤。預測欄位依屬性分為下列幾種,單價、組類別及補保資訊,預測的過程中先將品名進行萊文斯坦距離演算法的相似度計算,單價運用層次群集方法之華德法決定分群數,採用k-means執行分群,依據分析結果精進軍品料號系統功能。Traditional data mining technology focuses on descriptive analysis, analysis of information used to explain or explain. In recent years, data mining integrates a wide range of technologies, such as artificial intelligence, algorithms, statistics, database. Big Data is a synonym for the explosive growth of the amount of data. This paper adopts established data of military stock number system, and use data mining technology to get useful knowledge. We predict the possible input value of the field, this value can provide suggest answer, and decreased human negligence and error. Predicted value is divided into the following categories based on attribute of field, unit price, FSC, supply and maintenance information. In the process of predicting, we calculate the similarity of the item name based on the Levenshtein Distance algorithm. The number of clusters of unit price is determined by using the ward's method of hierarchical clustering method. The results of the clustering use the k-means algorithm. Improve the military stock number system based on analytical results.
關鍵詞 : 資料探勘(Datamining)、資料庫知識發現(Knowledge Discovery in Database)、群集分析(Cluster Analysis)、關聯規則(Association Rule)、決策樹(Decision Tree)