1 year vs 5 year PUMS input difference

replicahq / doppelganger

A Python package of tools to support population synthesizers

Apache License 2.0

165 stars 32 forks source link

1 year vs 5 year PUMS input difference #29

Open Shuake opened 6 years ago

Shuake commented 6 years ago

Hello,

I was experimenting with doppelganger using 1 year (2015) and 5 year (2011-2015) ACS PUMS records, and derived at very different sets of households/persons records. The 5-year PUMS record has larger sample size, however resulted in smaller sets of households/persons. I would appreciate any explanation as I might not be understanding/using the tool correctly.

You can find the inputs here, the outputs here, and the notebook here.

Thanks! Shuake

katbusch commented 6 years ago

Thanks for the detailed report! We haven't seen this and it will take a little time to investigate. I wonder if it has something to do with reusing marginals with different weights. @kaelgreco any ideas off the top of your head?