procedures for assessing consumer's preferences. Its main
objective is to measure individual's or population's preferences
on a set of (product) options that can be described by parameters
and their levels.
In this talk we consider choice based conjoint analysis, where
the preferences are assessed by choice tasks, i.e., by asking
respondents in a questionnaire to choose the most preferred
option from a small subset of options. The conjoint analysis
techniques that we consider here take the choices as input and
compute real number for every option, called the (scale) value
of the option. We approach the problem to compute scale values
by using two classes of methods: the first one is based on
statistical assumptions and the second one uses direct
regression. Both classes of methods can be used to estimate
scale values for either a population of respondents or for an
individual person. We introduce new computational methods for
both the approaches based on statistical distribution assumptions
and for the direct regression methods. We also conducted
extensive experimental tests to compare the different methods on
real and synthetic data for individuals and for populations of
individuals.