In Operations Research, sensitivity analysis describes the methods and tools used to study how the output of a model varies with changes in the input data. The input data may refer toparameters affecting the objective functions and/or constraints or to the structure of the problem. Depending on the problem and model, the output could refer to: * the optimal alternative and/or the optimal value, or, * a set of alternatives with a certain property. Some examples include the non-dominated set in a multi-objective optimization problem; the set of alternatives satisfying certain constraints in a classification problem; or the set of the, say, five best alternatives. Typical questions addressed within sensitivity analysis are whether a given optimal solution will remain as such if inputs are changed in a certain way, and,if not, which other alternatives may become optimal. Finding the most critical directions for changes in inputs that may affect the model output are also relevant sensitivity analysis issues, see French and Ríos Insua (2000), Saltelli et al (2000), and Saltelli et al. (2004) for reviews.
Autores: Redchuk, Andrés – Ríos Insúa, David