Testing decision services
You can test decision models, task models, and predictive models to verify that they
produce the results you expect. You can either create data sets for scenario simulation or
assertions for automated validation.
In Decision Designer, you test models by creating test data sets and running them. Two complimentary testing approaches are available:
- Data sets
- Data sets are sample information that is used to verify that a model behaves as expected in different situations. By creating and running data sets, you can simulate real business scenarios and confirm that the model produces the correct results. For example, you might test a loan approval model with various income levels and repayment periods to check that it makes appropriate decisions under different conditions. Running different test data sets helps demonstrate how changes in input values influence outcomes.
- Assertions
- Assertions are automated checks that confirm whether a model produces the expected result and alert you when something does not match. They help you verify specific values in the output, such as a customer's age, approval status, or address, by comparing the actual result with an expected outcome.
A test data set can contain a data set only, assertions only, or both. You create and manage test data sets from the Run tab available at the model level.
Warning: You cannot run models that contain errors.