DISCOVERING FUZZY CLASSIFICATION RULES FROM DATABASE BASED ON THE GENETIC ALGORITHM

Abstract

In this paper, we propose a method for discovering the fuzzy classification rules from a database based on the genetic algorithm. We would like to propose a method for choosing a set of threshold values which can help us to convert a fuzzy information table to a binary information table by using the genetic algorithm. We develop an algorithm for mining the fuzzy classification rules from binary information table and a fitness function which can measure the goodness of the discovered classification rules. A chromosome of threshold values has been used in genetic algorithm for selecting the appropriate set of thresholds for converting the fuzzy information table to binary information table with high goodness measure. We also propose an application to a supermarket database. In this application, we study the quantity of purchased items instead of 'to be purchased or-not to be purchased as the traditional approach did and mining the fuzzy classification rules from this database.