L. L. Chen, S. H. Yang, and J. W. Lin
National Taiwan University of Science and Technology
Under the current trend towards mass customization and agile manufacturing ( Anderson and Pine 1997; Pine 1992), it is becoming ever more important to understand what each consumer wants, and to assist consumers express their likes and dislikes. For studying the perceptions of products by target consumers, market researchers and designers often use perceptual maps for visualizing consumers' perception of different products and the critical factors affecting their decisions.
Product perceptual map can be constructed by using a number of methods, including multidimensional scaling (or MDS) [ Kruskal 1978; Schiffman 1981]. Based on proximities (or, conversely, distances) among the stimulus objects, MDS methods computes a (usually low dimensional) perceptual map such that each point in the map corresponds to a stimulus object, and distance between any two points match the proximity between the corresponding stimulus objects as much as possible.
By incorporating additional information about the rankings of stimulus objects according to a number of attributes, ideal vectors or ideal points, that represent the preferences of target. consumer groups, can be located in the perceptual map. From proximities among stimulus objects, it is also possible to conduct clustering analysis to partition the stimulus objects into clusters, each consisting of stimulus objects that are considered to be more similar than objects in other clusters.
Visualization of a perceptual map, where the image of a stimulus object is shown at the location of its corresponding point, is very useful for interpreting the perceptual space. Because the proximities among the stimulus objects are