View article

Using simulation to support making decisions imposes that models are thoroughly validated. We discuss widely adopted techniques for testing the validity of simulation results against real observations, as various techniques may be employed depending on the statistical features and the number of the available system/model data sets. Confidence-interval-based comparison provides certain advantages, but cannot be employed when a single system output data set is available. We propose an effective validation technique, performing confidence-interval-based comparison using a single system data set and multiple model data sets. The system data set is considered as a discrete signal and m-fold decimation is used, according to signal processing theory, fulfilling the essential requirements imposed to retain the statistical features of the original data set. We have evaluated both the effectiveness and the …