Check tax-unit benefit data (CPS) about taxdata HOT 8 CLOSED

I agree that it would be helpful to have an initial version of these tests up and running before we make UBI analysis available to the public on TB.

@andersonfrailey, could you add it to the grand task list and make the sure necessary responsibilities are accounted for?

.

from taxdata.

As for the tables themselves, everything from the C-TAM documentation would probably be valuable to start. There are certainly diagnostic histograms that could also be valuable. .

from taxdata.

@MattHJensen it wouldn't be difficult to generate those decile tables. Is the goal of these tests to make sure all the benefit numbers stay the same over time? What type of histograms would be good for diagnostic purpose? I assume we couldn't compare tax unit distribution with raw CPS distribution directly.

from taxdata.

There are a few goals

• Know exactly when and why results change.
• Provide up to date tables that summarize the database, allow for diagnostic review, and demonstrate the models capabilities.
• Provide up to date code for generating useful tables.

from taxdata.

What type of histograms would be good for diagnostic purpose?

For example, histograms of participants in a tax unit, like those histograms that helped in the recent SSI diagnostic effort.

from taxdata.

So if there's no objection or further discussion, I'm about to add a regression test script in taxdata repo. This script will generate 1) decile participation and benefit tables for each program, 2) participation and benefit totals, and 3) tabulation of tax-unit participation. Then the script will compare these tables with the existing versions or official stats if available.

@martinholmer @MattHJensen @hdoupe @andersonfrailey

from taxdata.

Thanks Amy. Sounds great.

from taxdata.

Closing this issue as PR #135 addressed the concerns here.

from taxdata.