The value of information in stochastic programming with partial information on probability distribution
Abstract: In many stochastic programming problems the value of the probability distribution may be partially known for the decision maker (DM) for many reasons (e.g. the information is too expansive). Stochastic program with linear partial information on probability distribution (SPI) is one of the models which deals with problems where we have an imprecise probability distribution. If the DM has the opportunity to get an additional information about the value of probability distribution, he may ask for the value of such information. In this paper we are interested in the value of perfect information and partial information on probability distribution in SPI problems. Some Bounds on such value are given based on the notion of upper and lower probability and the minimax theory.
Keywords: Stochastic programming, Partial linear information, Upper and lower probability, Minimax problems.