Mature market segmentation: a comparison of artificial neural networks and traditional methods

The research objectives are to examine neural networks, specifically Kohonen's self-organising maps (SOM), as an alternative to traditional statistical segmentation methods and to identify segments in the mature market which may direct its targeting.

Abstract

The need for in-depth knowledge of mature market segments and the need to overcome the limitations of using traditional methods to segment them motivate this study. The research objectives are (1) to examine neural networks, specifically Kohonen’s self-organising maps (SOM), as an alternative to traditional statistical segmentation methods (hierarchical and non-hierarchical cluster analysis) and (2) to identify segments in the mature market which may direct its targeting. The results show the superiority of non-hierarchical clustering and SOM over hierarchical clustering, and demonstrate their complementary nature. In addition, significant segments with particular characteristics are found in the mature market.