pysal / mgwr

Multiscale Geographically Weighted Regression (MGWR)

Home Page:https://mgwr.readthedocs.io/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Edit: Please delete - Binomial/Logistic Family not working

jagreen1 opened this issue · comments

EDIT: It appears that you should enter "Binomial()" without quotes for the code to work correctly. Please disregard this post.

I'm currently working on a GWR project using binary dependent data (0 or 1), and thus wish to use the binomial/logistic regression function. I have primarily been using the MGWR2.2 GUI Windows application, which allows you to define your model type as 'Binomial' without any problems.

I decided to redo the analysis with the actual python code, however the binomial/logistic model doesn't seem to work using pysal/mgwr 2.2 version.

When searching for the optimal bandwidth using 'sel_bw.py', defining the regression function family as either 'binomial' or 'logistic' results in the error "UnboundLocalError: cannot access local variable 'influ' where it is not associated with a value".
MGWR_Logistic_Error

Does anyone know if the binomial/logistic model works? What is going on in the GUI application that allows the binomial model to run?

MGWR_GUI_Binomial