Mining frequent pattern with Attribute Oriented Induction High Level Emerging Pattern (AOI-HEP)

摘要

This paper is extended version from previous paper which proposed AOI-HEP as novel data mining technique. This paper will explain how AOI-HEP mining technique can be used to mine frequent pattern. AOI-HEP is influenced by Attribute Oriented Induction (AOI) and Emerging Pattern (EP) mining techniques by applying AOI characteristic rule algorithm and improvement EP growth rate. The experiment used adult dataset from UCI machine learning repository with 48842 instances, run in 3 seconds and the instances were discriminated between government and non government concepts based on learning on workclass attribute. AOI-HEP mining interest for frequent pattern will be influenced by learning on their chosen attribute. The experiments showed that adult dataset which learn on workclass attribute had AOI-HEP mining interest for frequent pattern and there are four frequent patterns which have strong discrimination rule. Meanwhile, extended experiments upon adult dataset which learn on marital-status attribute showed there is no AOI-HEP mining interest for frequent pattern.