Dear Dinesh Babu,

Unfortunately MGET does not currently support that modeling technique. I am not very familiar with it so I’m not sure what to advise. The closest sounding thing I know of that is accessible from ArcGIS is the Geographically Weighted Regression tool from ArcGIS’s own Spatial Statistics Toolbox. I am not certain, but I suspect that tool is not non-parametric, however.

For general purpose predictive modeling, MGET provides GLM, GAM, tree models (a.k.a CART), and Random Forests. Linear Mixed Models are under development and mostly implemented. Regrettably, MGET does not currently provide any modeling tools that are traditionally used to explicitly address spatial autocorrelation, such as CAR, SAR, etc.

Best regards,

Jason

Hello,

I am Dinesh Babu, PhD student from Laval university, Canada. I have some questions on your tool before exploring it.

Is it possible to build non-parametric K-nearest neighbour regression statistical model using GIS shapefiles with your tool in ArcGIS?

Kindly reply me,

Thank you,

Dinesh Babu