Determine correlation coefficient for GWR

[StuartBrown]

Hi guys,

Just wondering if there is a way to calculate a localised correlation coefficient (local-r) for a GWR analysis rather than a localised R2. I'm asking because I have a univariate relationship that I believe varies from positively correlated to negatively correlated across my study region and of course the change in the relationship is lost when converting to R2. It can be easily done for an OLS model that is constant across space and I would like to know if it can be done for a GWR model?

Any help would be great.

Cheers,

Stu

[StuartBrown]

Hi Bill,

I do have a non-constant univariate relationship that I believe varies from a positive correlation to a negative correlation across my study area (27,000km^2). I could take the square root of R^2, but I would still not be any closer to knowing if it is a positive or negative relationship as all the results would be positive. If the GWR tool generating the results is calculating an R^2, it must be calculating an r value, I guess what I'm after would be a script that would output the r value with the results of the GWR. Any ideas?

Cheers,

Stu

[StuartBrown]

Never mind Bill, it just this second clicked in my head what you were saying.

Thanks a lot for the help!

Stu