Validation
Validation is checking if the model is a good model of the simulated phenomenon. A model has a degree of validity (Law and Kelton, 1991), it is not just valid or not valid. A model is valid to the extent that it adequately represents the syste mbeing modelled (Casti, 1997).
Introduction
Possible problems with validation:
- both model and system under analysis are likely to be stochastic
- a model might be able to produce plausible future predictions but may not be able to recreate known past system states.
- model could be correct but data from the real-world system may not.
- Many simulations are path dependant (i.e. the outcome of a simulation is dependent on the exact initial setup chosen) - history of a simulation is highly significant.
There could be problems at every single part and links of the following diagram:
How to
One type of validation: expert judgement
Outgoing relations
- Validation is part of Modelling Processes ABM
Incoming relations
- Pattern oriented modelling is a kind of Validation