Building Additive Utilities in the Presence of Non-Monotonic Preferences

Ordinal regression is used for modelling the problem of multiattribute choice in the presence of non-monotonic preferences. The main model presented in this paper estimates an additive utility function with non-monotonic partial utilities by analysing some ordinal inter-alternative information provided by the decision maker. Then the model is extended to fit in special conditions where intensity-of-preference information is available or concavity of the utility model is required.