1

I've got a polygon dataset with a few different variables.

I want to be able to see if these variables are spatially auto-correlated and then also run a Geographically Weighted Regression for the variables.

I feel like I did this in ArcGIS a long time ago and it was fine, but when I look at QGIS' tools for these things, I can only seem to input point datasets?

Is there something I am missing?

1
  • Please note that it is incorrect to define the spatial weights matrix (Wij) using centroid distance when your feature class are a polygons. Since polygons represent variable areas, you are changing the spatial process by coercing the data to points and then calculating distance. For polygon data the Wij matrix should be based on contingency or actual polygon distance (edge to edge). Oct 6, 2016 at 18:24

1 Answer 1

3

Well actually ArcMap uses points as well as the tool calculates the centroids first and then performs the Test.

(http://edndoc.esri.com/arcobjects/9.2/net/shared/geoprocessing/spatial_statistics_tools/spatial_autocorrelation_morans_i_spatial_statistics_.htm)

That means you can just calculate the centroids in QGIS and use the tools you found. There are two ways:

One is the Extension "Real Centroids" which gives you already the point-geometry as output and also displaces the centroid if it is not within the polygon (what sometimes happens when you just calculate the coordinates in the attribute table)

Or you can calculate the coordinates with the field calculator.

x_min(centroid($geometry))
1

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.