eli knaap

copiolt thinks the failure is a caching issue (apparently), so i'd guess it will pass if re-run on its own

looks like maybe some other parts of the stack are moving at the moment, but let me know if there's anything I can do to help move this forward :)

no sweat! thanks for the comments right there with you and similarly spread thin :). Static typing is pretty extraneous in most of my projects so not very practiced; will...

@kylebarron I think this is ready for another review, though it looks like i also pulled in a bunch of other stuff by accident. Maybe i should close this and...

> In particular, there should be virtually no `type: ignore` statements. Everything is typed now, so the only linter complaints are the four on line 55, all of which are...

the trouble here is that vector-based binary dasymetric (as opposed to something like [multiclass](https://doi.org/10.1016/j.compenvurbsys.2004.07.001)) is *just* an overlay operation. Whether you have ancillary data that defines the inclusion or exclusion...

[here's](https://gist.github.com/knaaptime/f2cf33e25e7182d727073798029e195e) a demo notebook that gives us some code to discuss So, which of these workflows is preferable? The first: ```python ancillary = ancillary.query(subset) # refine ancillary data source =...

ok, i'm open to it. >From a very pragmatic perspective, the second. You can avoid calling geopandas.overlay which does not only trims the geometry but also spatially joins the attributes....

> but the dasymetric mapping in Eicher & Brewer retains the area, but maps it with zero population. that's true, but there's no need to retain the geometries, especially because...

>It's also not quite clear to me how an inclusion layer would work in Lee's approach? Some sort of groupby? i guess you just swap the zero-out, dur :P