pysal/inequality vs. pysal/segregation?

[jGaboardi]

At some point we should take a deep dive to figure out the overlap between inequality and segregation, and determine how much/if any of inequality can be consumed by segregation (or vice-versa). Or even if inequality should be entirely cannibalized by segregation.

[martinfleis]

I like that inequality is very lean package. momepy uses Theil and Gini and I would rather not depend on whole segregation here with rvlib et al as dependencies. But I can easily move those simple formulas over and not depend on any of those.

From a wider perspective, it makes sense to merge this stuff with something else, it is tiny and the last release was 4 years ago, so not even active.

[jGaboardi]

To be clear & transparent, I think that the most reasonable path forward is for inequality to remain its own package. I was simply throwing out some possible ideas.

As for the maintenance side, I have been spending time cleaning inequality up over the last month and would very much like to get a fresh release out to be included in the next PySAL meta release.

[knaaptime]

serge and i have talked about this once or twice before. I tend to agree a merger makes sense

if it would help, I think its probably worthwhile to make rvlib an optional dependency (and fallback to the slower sampler otherwise), so segregation's requirement footprint should become pretty standard (i think quilt can go too)

[martinfleis]

make rvlib an optional dependency

That would be good!

[sjsrey]

I think inequality and segregation should remain separate.

There is a bunch of inequality enhancements I have code for but have not pushed up, related to an nsf project that just is winding down.

[jGaboardi]

The big boss has spoken. I'll close this out.