Rough Clustering of Destination Image Data Using an Evolutionary Algorithm

Abstract

This article describes a hybrid computational intelligence technique, an evolutionary algorithm based rough clustering algorithm. The technique of cluster analysis is fundamental in traditional data analysis and data mining, and is used to group similar objects together. Many clustering methods have been identified, including the commonly used k-means approach, which is dependent on initial starting points and requires the number of clusters to be specified in advance. The rough clustering algorithm described in this article is able to overcome these limitations. Rough clusters are defined in a similar manner to the rough set concept developed by Pawlak—that is, using a lower and upper approximation. This allows for multiple cluster membership for objects in the data set. The lower approximation of a rough cluster contains objects that only belong to that cluster. The upper approximation of a rough cluster contains objects that belong to more than one cluster. The article describes the template, the data structure used to describe rough clusters. It also provides an overview of the evolutionary algorithm used to develop viable cluster solutions, consisting of an optimal number of templates, which provide easily interpreted descriptions of the clusters. This evolutionary algorithm based rough clustering algorithm was tested on a large data set of perceptions of city destination image attributes. Some preliminary results are presented.