Gestión y política pública
versão impressa ISSN 1405-1079
FERNANDEZ GONZALEZ, Eduardo; LOPEZ CERVANTES, Edy; NAVARRO CASTILLO, Jorge e VEGA LOPEZ, Inés. Application of Multi-Objective Metaheuristics to Public Portfolio Selection Through Multidimensional Modelling of Social Return. Gest. polít. pública [online]. 2011, vol.20, n.2, pp. 381-432. ISSN 1405-1079.
This paper is devoted to the problem of allocating public funds to competing programs, projects or policies. Under a subjective approach, we apply multi-criteria analysis to define the concept of highest social portfolio return. This concept is modeled as the best compromise solution of a 0-1 multi-objective optimization problem. Often powerful metaheuristics are required for solving real size problems, but with many objectives evolutionary algorithms are also inefficient. Such difficulty is overcome using a new method which incorporates decision-maker's preferences to the evolutionary search. We used a model based on a binary fuzzy outranking relation provided by ELECTREE methods for multi-criteria decision making. The proposal is examined through real size problems, in which good solutions are reached; examples illustrating the case of basic research project portfolios and social action program portfolios are shown here.
Palavras-chave : public portfolio selection; multi-objective optimization; evolutionary algorithm; fuzzy outranking relations.