andrew.pearce.carter

Geographically Weighted Regression and strange global R2

Discussion created by andrew.pearce.carter on Dec 11, 2013
This is definitely a little confusing; I just started using GWR and the values I'm getting for local R2 look accurate (I also ran a LISA analysis and the high R2 are more or less located where both the dependent and independent variables are highly clustered), but the global adjusted R2 is way too high (.29). 

.29 is the value of the highest local adjusted R2 but there aren't that many locations approaching that number, and when I separately ran an Exploratory Regression in ArcGIS and a regression in Excel, both gave me a far more reasonable-sounding adjusted R2 of .12.

So am I missing something here?  The global adjusted R2 is just supposed to be what you get with no geographical weighting, right?

Outcomes