Positioning Analysis with Self-Organizing Maps

Abstract

Various techniques of multivariate analysis are applied to locating hotel chains' positions on perceptual maps based on consumers' views of those chains. Measuring and accounting for the many attributes used to develop positions is intensive and usually the analysis is segment-specific. Instead of complex methods that are hard for managers to apply, the use of self-organizing maps may be appropriate. The self-organizing map is a computer-based application of neural-network models that treat data as nodes with connections. Data are sorted to arbitrarily positioned nodes that represent a map of positions. In the simplest approach, the result is that the positions of hotels that fall on the same node are not different, while those that fall on adjacent nodes are different. In actuality, a hotel's position falls on one node sometimes and other nodes other times, depending on the market segment. The higher the frequency of a hotel's being located on a given node, the more focused its image. A hotel that finds itself widely distributed on many modes has a fuzzy image. An example of luxury hotels in Vienna demonstrates the hotels' positions for different guest segments.