the process of determining how well one system replicates properties of some other system or, more generally, any comparison between the representation of a system and specified criteria. The validation of an operating
model cannot be separated from the purpose for which it is designed and used.
Validation is the process of increasing the confidence that the outputs of the model conform to reality in the required range. In some cases, the model's output can be compared to data from historical sources or from an experiment conducted for validation. A model can never be completely validated. We can never prove that its results conform to reality in all respects. It can only be invalidated. Predictive models be validated only by judgment, since a model may fit past data well without having good predictive qualities. (IIASA)
