Use a test to identify lost facts & assert some stuff about them

[jdangerx]

One day, we will be able to run this test in CI and it will tell us when there are too many lost facts.

For now, we're just using the test as a simple script runner that helps us explore the lost facts.

TODO:

• get more statistics out (e.g. including the context dimensions in the statistics)
• assert across all instances

We might want to assert something like "in every instance, the % of lost facts named order_number is less than 0.5% of the total facts".

Or, we might want to assert something like "in every instance, the % of lost facts at all is less than 1% of the total facts."

Or, "across all instances at once, the % of lost facts is < X%"

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[jdangerx]

Closing in favor of catalyst-cooperative/pudl#2774