Study on a Data Warehouse Mining Oriented Fuzzy Association Rule Mining Algorithm

Abstract

This paper concentrate the problem of fuzzy association rule mining for data warehouse mining. The aims of our algorithm are to obtain the frequent fuzzy grids, and then mining useful fuzzy association rules. Next, some definitions are provided to describe the theoretical basis of this paper. Afterwards, the framework of the proposed association rule mining algorithm consists of four parts: (1) Mining useful information from the website database and Web log file, (2) Constructing the basic data source using hierarchy trees and fuzzy sets tables, (3) Generating the large scale item sets from the initial data in data warehouse based on the fuzzy association rules mining algorithm, and (4) Obtaining fuzzy association rules by the fuzzy rule generating algorithm using the data extracted from the former three steps. Furthermore, the fuzzy rule generating algorithm is illustrated, which is made up of four steps. Finally, the adult dataset is utilized to make performance, and this dataset is used to judge whether a person makes over 50K a year. Furthermore, three kinds of probability distributions are exploited, such as random probability distribution, high probability distribution and low probability distribution. The experimental results show the effectiveness of the proposed algorithm.