Closed pedrovma closed 3 years ago
This looks good!
I do worry about duplication of spopt.region.skater
though... I wonder if it's possible to fully re-use the Skater class & just use a different "score" function for Skater_Reg
? If you remove the "attribute_kernel" step there on line 77 and pack up "data" as you have with the regression dictionary, I think you could get away with only re-writing the score function and leaving the core of .fit()
?
Something for the future to investigate, if necessary. Definitely don't want to hold this up, as it's really cool!
Adding Skater Regression to allow for the estimation of endogenous regimes regression, based on: https://www.researchgate.net/publication/353411566_Endogenous_Spatial_Regimes
The PR also adds a notebook to show Skater_reg usage.