A property score, like a metric, should not depend on the index of the real and synthetic data. It should only depend on their column values. It would be nice to have some integration tests that check that this is the case for all our current properties (single and multi-table) as well as for future ones.
Expected behavior
Write a simple integration test where real and synthetic data are the same and only differ by their indexes. Check that the property score is 1.0. Apply this test to all the properties (keeping in mind that it should work for future ones also). In the case of some property scores not being 1.0:
Investigate if an easy fix could solve the issue
Otherwise, an easy fix would be to copy the data and reset the index at the beginning of any reports.
Problem Description
A property score, like a metric, should not depend on the index of the real and synthetic data. It should only depend on their column values. It would be nice to have some integration tests that check that this is the case for all our current properties (single and multi-table) as well as for future ones.
Expected behavior
Write a simple integration test where real and synthetic data are the same and only differ by their indexes. Check that the property score is 1.0. Apply this test to all the properties (keeping in mind that it should work for future ones also). In the case of some property scores not being 1.0: