Closed rwesterh closed 2 months ago
@rwesterh Thanks! Could I ask whether you considered rather creating these functions as say methods for a fitted "localmoran"
object, perhaps like the hotspot
methods in R/hotspotmaps.R
(perhaps moran.plot
itself ought to be a method?)? I think this might reduce some code duplication. In addition, the localmoran
object could be enhanced to add columns you need. I see a problem in that the ZIi
from the samples in column 4 is not robust to variables that are skewed, as it is based on the mean and var of the samples. This probably means looking at the skewness and kurtosis reported in the object, and warning the user if the p-values from mean and var of the samples diverge from the p-values from ranking the samples. This also impinges on https://github.com/pysal/esda/issues/199, from https://github.com/pysal/esda/issues/199#issuecomment-1544126890. You may also be interested in doi = {10.1080/24694452.2024.2326541} if you have not seen it yet, also maybe doi = {10.1111/gean.12390}. Quite a lot of interesting suggestions coming forward now!
I have added two new versions of the Moran scatterplot: the Moran drop plot (for visualising p values) and the Moran seismogram (for visual inspection of local critical values/potential spatial weights misspecifications).