Model verification and validatioN

All models require to be validated and verified.

  • Verification is the process of making sure your conceptual model matches your implemented model.
  • Validation is the process of making sure your implemented model corresponds to the real world.

Verification

  • Communication is essential to verification
    • Often the model developer and the model author are not the same person.
    • Reducing / eliminating the gap between developer and author improves verification.
  • We need a common language to describe conceptual models that is recognized by both authors and developers.

Validation

Ensuring that the implemented model corresponds to reality.

  • All models are simplifications of reality.
  • “All models are wrong but some are useful.” – George E. P. Box.

Types of validation

  • Macro-Validation: Comparing aggregate results at many different levels.
    • Are the patterns of adoption similar?
  • Micro-Validation:
    • Comparing individual rules: Do consumers use the same information when deciding whether or not to adopt?
    • Comparing individual properties: Are the properties similar?
  • Face Validation: Do the general ideas about behavior and properties compare to real-world phenomena?
  • Empirical Validation: Does data from the model correspond to real-world data?

Leave a Comment