A novel associative classification algorithm: A combination of LAC and CMAR with new measure of weighted effect of each rule group

Author

Hao, Pei-Yi ; Chen, Yu-de

Author_Institution

Dept. of Inf. Manage., Nat. Kaohsiung Univ. of Appl. Sci., Kaohsiung, Taiwan

Volume

2

fYear

2011

fDate

10-13 July 2011

Firstpage

891

Lastpage

896

Abstract

In recent, Association Classification not only has widely adopted but also has performed well in data mining. The literatures have been argued that the small disjunction and using multiple class-association rules have significant effect on classification accuracy. This paper is based on CMAR (Classification based on Multiple Class-Association Rules) and Adriano Veloso proposed Lazy Associative Classifier algorithm for Small Disjunction mining. In addition, we collocate with a new weight calculation method in our algorithm to solve weight bias problem of CMAR. This paper uses UCI 26 data set for experiment on our proposed algorithm. The finally results convincingly demonstrated that our proposed algorithm is high accuracy.

Keywords

data mining; pattern classification; CMAR; LAC; UCI 26 data set; associative classification algorithm; classification accuracy; data mining; lazy associative classifier algorithm; multiple class association rules; small disjunction mining; Accuracy; Classification algorithms; Filtering algorithms; Machine learning; Machine learning algorithms; Testing; Training; Association Rule; Associative Classification; CMAR; Data mining; LAC; Multiple Rules;

fLanguage

English

Publisher

ieee

Conference_Titel

Machine Learning and Cybernetics (ICMLC), 2011 International Conference on

Conference_Location

Guilin

ISSN

2160-133X

Print_ISBN

978-1-4577-0305-8

Type

conf

DOI

10.1109/ICMLC.2011.6016766

Filename

6016766