A recursive algorithm for volume-based sensitivity analysis of linear decision models

Sensitivity measures for linear models using volume in parameter space have been proposed by various researchers as early as Starr’s domain criterion in 1962. These concepts have distinct advantages over more widely used distance-based approaches, but require complex computations for determining the volumes of polyhedra in n-dimensional space. In this article, we present a simplex-based recursive algorithm which, in contrast with more general algorithms, exploits the structure of the underlying linear model. The running time of this algorithm is analyzed both in terms of the number of alternatives and attributes of the problem.