Feature idea: ETL characterizations

[katy-sadowski]

Could we output summary stats like the delta in patient count between the source database and the CDM?

[lawrenceadams]

This is a great idea - do you have an idea of how you'd want to do this practically? I've seen it done in the past as dbt tests where you check distinct counts from the source and compare to your final table etc?

[katy-sadowski]

This was something @fdefalco asked about at the demo yesterday - haven't given it much thought yet but the idea is supplementing the documentation with stats that allow users to understand the impact the OMOP transformation had on the actual contents of the dataset.

[lawrenceadams]

Ahhh I see! Stuff like column level lineage etc is nice for this, easier said than done - but a good thing to think about

[lawrenceadams]

Something like this?