search

pysal / gwr

This is a repository for the geographically-weighted regression submodule of the Python Spatial Analysis Library
BSD 2-Clause "Simplified" License
16 stars 16 forks source link

Local mc #22

Closed TaylorOshan closed 6 years ago

TaylorOshan commented 6 years ago

First pass at add local measure of multicollinearity, including local correlation coefficients, local VIF, local condition number and local variance decomposition proportion. Not 100% sure about the delivery...one method for all four of the indicators, but several of them require similar inputs or are a derivative of each other so this seemed the simplest way. Also for tests, just manually added some cvs files with known values validated against GWmodel, because not sure where new test-related files should go with refactor. There is still the example repository in libpysal, but does that mean we would need to make a PR to libpysal anytime we want to update/add any example, even if it not relevant to anything else?

TaylorOshan commented 6 years ago

Good catch, those checkpoints are gone now

Ziqi-Li commented 6 years ago

Just did a quick test using election dataset against GWModel and all good.

TaylorOshan commented 6 years ago

@ljwolf @Ziqi-Li Y'all cool with where this is at? Would be good to get in merged in, as I'm not experimenting with MGWR extensions.