Using Statistical Models for Mathematical Programming Sensitivity Analysis

Abstract

When exploring the potential payoffs from energy R & D, it is necessary to perform benefit-cost calculations over a wide range of future scenarios. Repeated runs of large-scale mathematical programming models are often required. This paper describes how simple statistical models might be used to eliminate at least some of the costly computer runs. The approach is illustrated in the context of a recent study on the effect of breeder timing on electricity generating costs.