A DATA mining Procedure Using Neural Network- Self Organization Map and Rough Set to Discover Association Rules

Abstract

Modern enterprises are usually equipped with database and data warehouse for data storage and analysis, but they are of less use without insight and intellectual analysis. Data mining techniques are popular and often used to find knowledge from enterprise data store and thus support to make better decisions. Data mining by automatic or semiautomatic exploration and analysis on large amount of data item set in a database can discover significant patterns and rules underlying the database. However, for large non-homogeneous data item set, direct extraction of rules with traditional data mining techniques may be useless or meaningless. The purpose of the paper is to propose a data mining procedure for database with large amount of transaction records. The data mining procedure consists of two methods. One is a neural network, Self-Organization Map (SOM), which performs an affinity grouping tasks on large amount of database records. The other is rough set theory, which extracts association rules for each homogeneous cluster of data records. An implemented system was applied to a sample from purchase records of a database for illustration.