Possible process for bringing TaxData puf.csv closer to IRS published data and forecasts of same

[donboyd5]

Tagging people who I think may be interested in this topic but please add others if I missed someone:
@andersonfrailey @MattHJensen @rickecon @jdebacker @martinholmer

Over the last year or so I have stumbled across several instances in which TaxData puf.csv as grown and weighted is not as close to IRS totals or potential forecasted targets as might be ideal. @jdebacker recently sensibly closed several issues I had raised in Tax-Calculator as they were more appropriate to TaxData. Rather than reopen them here, let me just list them: TC2503, TC2505, and TC2506. In addition, I've opened #388, #398, and #399 in TaxData.

I believe these issues are highly resolvable. However, it might make sense to address them en masse through some enhancements to the data and process for targeting.

I realize that would take resources to allow more person-hours of work on TaxData - possibly resources that the TaxData project doesn't have. Or perhaps it can be done within existing resources - I'm sure @andersonfrailey would have the best sense of that; it's not something I know. But I think it would add a lot of value to a project that is both amazing and important. Thus, I'd like to suggest two topics of discussion here:

• What enhancements to the targeting data and process might make sense and how much effort would that take?
• If it requires raising money, how to go about it?

If there seems to be interest in this topic, I'll come back with some thoughts on both. And if there's a better way to have a discussion about this, please let me know.

[andersonfrailey]

I would definitely be open to enhancing the overall targeting process. I'm going to need to take some time to think through the issues you've listed and come up with potential solutions. In the meantime I think this issue is a good spot for anyone who has ideas to post them.