I am estimating a gravity model of migration on cross-sectional data. The Moran I statistic indicates a positive and significant spatial autocorrelation in the residuals of the non-spatial model, and the Lagrange Multiplier test points to the Spatial Autoregressive (SAR) model as the preferred specification.

While I have no issue fitting a linear SAR, it does not accommodate the very large number of zeroes (> 90%) in my dependent variable. This clearly point to a Poisson process.

In short, I am having trouble fitting the SAR Poisson model. I had two questions:

  1. Is there a method to run a SAR Poisson GLM in R? (I searched quite a bit before posting here)
  2. If answer (1) is no, I should at least use a spatially filtered Poisson GLM. Yet, both SptatialFiltering() and ME() methods crash even using a very simple connectivity structure (symmetric knn = 5). I mean that it did not give any message error but RStudio simply "lost the connection with the R session". I suspect this is due to the large number of observation (278 784). Any tip to increase computational efficiency?

I am aware this question focuses on programming but I posted it on CV rather than SO because it require a statistician rather than a programmer.

  • $\begingroup$ Did you try install.packages("sos", dep = TRUE); library(sos); findFn("SAR Poisson")? $\endgroup$ – user81847 Feb 1 '16 at 9:52
  • $\begingroup$ See also stat.ethz.ch/pipermail/r-sig-geo/2016-February/023977.html in case you get answer there. $\endgroup$ – user81847 Feb 1 '16 at 10:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.