Join GitHub today
G_Local: EG_sim and seG_sim are scalar #53
In Moran_Local(), EI_sim and seI_sim are nx1 arrays, so differentiating the outcome by observation. In G_Local, EG_sim and seG_sim are scalar, in some cases that I've looked at narrowing the z_sim distribution compared to Zs, the analytical distribution.
Was this intended, or should G_Local be reviewed around line 376 in esda/getisord.py? Are there a vectorised sum() and std() methods to apply to an array row-wise?
Thanks for spotting this. It is a bug and was not intentional as the treatment in G_Local() should mirror that of Moran_Local() in the treatment of the z_sim distributions. The original code seems to be a case of a cut and paste error.
Will be submitting a fix shortly.