關聯規則探勘是資料探勘領域其中一種最廣為人之的探勘方法，其主要內容是在一組交易資料中計算不同商品同時被購買的頻率，進而找出這些共同被購買之關係中的規則。另一方面關聯規則在解決分類問題之應用層面亦已行之有年（關聯式分類）。然而一旦分類規則產生出來，其缺乏組織反而造成閱讀與理解上的缺陷。為了解決此點，因此本文提出從關聯規則集中摘要以及建立決策樹的構想與具體作法。期望結合兩者優點來建立分類模型。就分類模型而言，此方法連結關聯式分類與決策樹二者之優點：相較於前者更加具理解力、有組織，精簡、容易使用的分類模型；相較於後者分類正確度亦比傳統C4.5建立決策樹方式來的更為精確。 Association rule mining is one of the most popular areas in data mining. It is to discover items that co-occur frequently within a set of transactions, and to discover rules based on these co-occurrence relations. Association rules have been adopted into classification problem for years (associative classification). However, once rules have been generated, their lacking of organization causes readability problem, i.e., it is difficult for user to analyze them and understand the domain. To resolve this weakness, our work presented two algorithms that can use decision tree to summarize associative classification rules. As a classification model, it connects the advantages of both associative classification and decision tree. On one hand, it is a more readable, compact, well-organized form and easier to use when compared to associative classification. On the other hand, it is more accurate than traditional TDIDT (abbreviated from Top-Down Induction of Decision Trees) classification algorithm.
