Colors are not fixed is LISA maps

[serenini]

When I create a map that does have regions in the 4 classes, the colors are ok (Light blue used to LH):

But when I create a map that does not have regions in each of the 4 categories, the colors are changed (orange used to LH):

How can I fix the colors to compare the same map in different times?

[sjsrey]

@alvinegro can you add a code snippet so we can help debug?

[serenini]

@sjsrey I have a dictionary called "separado_rgi" with dfs from 2002 to 2017.
I want to create LISA maps from the variable "rendimento" to compare each year.
I created a LISA map for each year:

import pysal as ps
from pysal.explore.esda import Moran, Moran_Local

w = Queen.from_shapefile('dados/RG2017_rgi.shp')
lisa =Moran_Local(separado_rgi[2003].rendimento, w)
lisa_cluster(lisa, separado_rgi[2003], p=0.05, legend=True);

The first code above, generated this map:

In 2003, there was significant regions in all the 4 categories, but, in some years (e.g. 2006) some categories does not have any region with significance level. So when I created the map, the colors related to each category was changed (now, LH is orange, istead of lightblue):
lisa =Moran_Local(separado_rgi[2006].rendimento, w)
lisa_cluster(lisa, separado_rgi[2006], p=0.05, legend=True);

I would like that the colors be the same for all over the years, so it would be possible to compare the evolution of the clusters.

[sjsrey]

I can't reproduce this on other examples, so a couple of options:
In the call:
lisa_cluster(lisa, separado_rgi[2003], p=0.05, legend=True)
the second argument should be a geodataframe. Can you try changing this to see if the result is what you are looking for?

If that doesn't work, can you send your data so I can debug locally?

[slumnitz]

Hello @alvinegro, which splot release are you using? already v1.1.1? If not then this could be fixed by using the most up to date version. If it is still a problem, I know where the bug is. Just let me know.

[sjsrey]

I can't reproduce this on other examples, so a couple of options:
In the call:
lisa_cluster(lisa, separado_rgi[2003], p=0.05, legend=True)
the second argument should be a geodataframe. Can you try changing this to see if the result is what you are looking for?

If that doesn't work, can you send your data so I can debug locally?

I misread your earlier response, and now see you are passing a geodataframe in the second argument. So, I can check things locally if you are able to share the data.

[serenini]

@slumnitz how can I check or update the version of the splot release?

[serenini]

@sjsrey I sent the data to your e-mail

[slumnitz]

if you used pip for the installation: pip install —upgrade splot

if you used conda: conda update splot

[serenini]

@slumnitz I updated the splot to version 1.1.1 but the problem persists

[slumnitz]

hello @alvinegro, I am sorry, but I cannot reproduce your error. Sorting the right label with the right colour works for all of my datasets:

I would highly recommend double checking your splot version number again. Potentially reinstalling splot if the update did not work for you. Note: the new splot version is only included in the newest PySAL release.

Otherwise, feel free to also share your data with me and I can see if I can get it working or get it debuged.

[serenini]

Hello @slumnitz ! I noticed that you already solved this problem at #73 , so I coppied the code of lisa_cluster and mask_local_auto on another .py file, so I imported lisa_cluter from it and worked.
I will try to reinstall splot.
Anyway, thank you very much for your help!