Multiobjective decision making under certainty

Abstract: A decision maker often has more than one goal to consider. On a macro-economic level many factors inflation, foreign trade, economic growth, employment and various distribution effects are balanced against one another. On a micro-economic level management has to consider the way in which decisions affect profit, liquidity, sales and other goals. In this dissertation various methods are discussed for helping the decision maker to weigh different goals against one another in a systematic way. Two main approaches one traditional and one more recent are discussed. The first approach is based on the assumption that it is possible to estimate a globally valid model of the decision maker’s utility function. Some methods and models used in estimating utility functions are discussed. Many authors claim that linear utility models are good models, particular when compared with non-linear ones. The relevance of the customary measure of "goodness"and of the usual non-linear models is questioned in this dissertation and as a result of the analysis a new model is suggested. The second approach starts from the assumption that good decisions can be achieved with the help of man-machine interactive methods. Such methods have several advantages. There is no need to generate explicit utility models, which are supposed to be globally valid. The decision maker has to provide local preference information only. His efforts are immediately rewarded in terms of continuously better alternatives: he finds out what can be achieved and he can interrupt the process when he wants to. The advantages and disadvantages of interactive methods are discussed. A new method is presented. In a comparative simulation experiment this method appears to be efficient and robust even in the presence of error in the information provided by a simulated decision maker.