Which Sensitivity Analysis Method Should I Use for My Agent-Based Model?
Résumé
Sensitivity analysis (SA) is usually an overlooked steps in agent based modeling exploration (according to [3], 95% of sim-
ulation models reviewed do not use statistical validation techniques). This lack of scrutiny includes: to determine arbitrarily
the number of replicates to flatten stochastic processes impact on simulation outcome, fixed arbitrary parameter values to
lower the dimension of the problem to explore, empirical scenario search to find points of interest in the parameter space,
weak evaluation and validation of simulation outcomes and parameter contribution; among other issues. Potential explanation
includes lack of knowledge, duration of the procedure, computational power required, lack of generic techniques, algorithmic
complexity and accessibility of methods [1].
Origine | Fichiers produits par l'(les) auteur(s) |
---|