Abstract:

This paper introduces - necessarily briefly - OPAS, a system to optimise products and assortments. Basically, two parts can be distinguished. The first part consists of all models related to data collection and conjoint measurement, resulting in an array of values (called utilities) for each attribute level. Starting with traditional conjoint measurement, problems for applications are summarised. For a significant part of these, solutions have been formulated and the implications for the models are shortly indicated. Thus, it is possible to include a large number of attributes and to apply this system to taste-testing. The second part contains models for prediction (including a set of probabilistic choice models) and the routines for optimisation. Since the latter can include constraints (e.g. costs, number of products) a large amount of practically relevant questions can be answered. This is illustrated by a sample of the situations where OPAS has been successfully applied, which is followed by a concluding section on limitations and future extensions.