I haven't been able to find out how to blend correlations across states, but I have seen how we can add a region GP to the state GP, which kinda does the same thing. But we'll need to define region. Past models group it geographically (NE, SE, MW, NW, SW, etc.), but I think we can do better. If we group by demographics (age, income, race, religion prop, etc.) I think we can it would be better.
Probably will work as a side module. And we can verify our groupings by checking to make sure WI and MI are in the same group.
I haven't been able to find out how to blend correlations across states, but I have seen how we can add a region GP to the state GP, which kinda does the same thing. But we'll need to define region. Past models group it geographically (NE, SE, MW, NW, SW, etc.), but I think we can do better. If we group by demographics (age, income, race, religion prop, etc.) I think we can it would be better.
Probably will work as a side module. And we can verify our groupings by checking to make sure WI and MI are in the same group.