For the analysis of preference ratings which are presented in the form of s
everal data sets, each data set being associated with a consumer, we run a
method of analysis called Common Components and Specific Weights Analysis.
This method makes it possible to determine the underlying dimensions to the
preferences expressed by the consumers. Moreover, it computes for each con
sumer and for each underlying dimension a weight that reflects the importan
ce that this consumer attaches to the dimension being considered. In a subs
equent stage, we performed a clustering of the consumers on the basis of th
eir agreement upon the preference dimensions. (C) 2001 Elsevier Science Ltd
. All rights reserved.