I want to run a LISA with a time series spatially lagged variable using PYSAL. I see that LISA will allow the definition of y & w but not Wy(t-k).

Any thoughts how I can get the lagged variable into PYSAL?

I also notice that PySal has a BV Morans but I can find nothing for a BV LISA.


I believe that "pysal.weights.spatial_lag.lag_spatial" and "pysal.esda.moran.Moran_BV_matrix" are what you are looking for. There is documentation here: http://pysal.geodacenter.org/1.4/library/index.html

  • Thanks. Neither runs the LISA however. The former will create my lagged variable and the latter will run Morans Matrix but I really need a decomposed local statistic. Something like: class pysal.esda.moran.Moran_BV(x, y, w, transformation='r', permutations=999) where I can plug in x, y, w but get returned not one singular value but a local indicator of SA. Thoughts? – Rich H. Nov 1 '12 at 17:51
  • The help seems to indicate that the result is local "(i, j) is the key for the pair of variables, values are the Moran_BV object". Although, specifics are sparse and there is no indication if this is locally decomposed and integrates to the global. Perhaps you should contact the developers for clarification. Is there a particular reason that you are wanting to stick to PySAL? The bivariate LISA is available in GeoDa. Personally, I would conduct this analysis in R, which has the capacity for creating temporal spatial lags and the bivariate LISA. Sorry I could not be of more help! – Jeffrey Evans Nov 1 '12 at 18:23
  • No. You have been! I have everything in Python and would like to stick to it. I know about the functionality in GeoDa, which surprises more that I cannot find similar functionality in PySAL. The BV Matrix seems to give results for i,j in a set a variables, which is actually very cool but it doesnt seem to give it as an array. UGH! – Rich H. Nov 1 '12 at 19:03

Your Answer

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

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