A Study of Constructing an Associative Classification Tree
| Autor: | Chien-kang Wang, 王健剛 |
|---|---|
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 97 Associative classification is a branch of data mining algorithms that classifies data by association rules. It’s also a special case of association rule mining. Associative Classification has better accuracy than traditional ID3, C4.5, but it will reduce the readability when it has huge amount of rules. Therefore, A study of ACT(Associative Classification Tree) algorithm was proposed. It builds a decision tree by many classification association rules which combines the advantages of associative classification and decision tree. The structure of associative classification tree is much simple because it only keeps some important rules. Also, it increases the readability of associative classification. However, one of the ACT algorithms is effected by “base rules” of a node of the tree which cannot select any better splitting attribute, and decreases the accuracy of the tree. Therefore, our purpose in this paper is to implement the ACT algorithm, and try to improve it. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|