Facetgrid (multiplots)

[sarahclaude]

I have done some work on gridmap and scattermap to produce facetgrids (multiplots) with the option;
plot_kw = {'col': 'coordinate', 'row': "coordinate"}.

Gridmap seems to be working fine and I am still working on scattermap (the marker sizes can't be send directly to xr.plot.scatter since the results given back are weird).

Once done I will start hatchmap. I think the options to add multiple xr.Datasets/DataArray via a dictionary should be removed since it will be confusing if passed with plot_kw = {'col': 'coordinate'}. This option is also absent in the two other maps options.

[juliettelavoie]

as-tu réussi à faire une gridmap et scattermap ensemble sur une figure? Je pense que ça va être plus compliquer qu'on pensait.. https://stackoverflow.com/questions/44158276/combine-different-seaborn-facet-grids-into-single-plot

[juliettelavoie]

Maybe I would put the y label of rows in the left side of the facetgrid instead of the rigth, next to the colorbar.. Or I would put the colorbar at the bottom. It just feels a bit crowded to me.

[vindelico]

as-tu réussi à faire une gridmap et scattermap ensemble sur une figure? Je pense que ça va être plus compliquer qu'on pensait.. https://stackoverflow.com/questions/44158276/combine-different-seaborn-facet-grids-into-single-plot

The link suggests you are asking about having a facetgrid with gridmap(s) and scattermap(s) alongside each other, i.e. different tiles of the facetgrid have either gridmap(s) or scattermap(s).
A different way of combining them is my use case for stacking facetgrids of gridmaps and of scattermaps, so that the scattermaps are superposed over the gridmaps, in other words: dots on maps. I will look into that.

[sarahclaude]

as-tu réussi à faire une gridmap et scattermap ensemble sur une figure? Je pense que ça va être plus compliquer qu'on pensait.. https://stackoverflow.com/questions/44158276/combine-different-seaborn-facet-grids-into-single-plot

The link suggests you are asking about having a facetgrid with gridmap(s) and scattermap(s) alongside each other, i.e. different tiles of the facetgrid have either gridmap(s) or scattermap(s). A different way of combining them is my use case for stacking facetgrids of gridmaps and of scattermaps, so that the scattermaps are superposed over the gridmaps, in other words: dots on maps. I will look into that.

I think we should try these functions instead since xarray creates its own facetgrid object with its own methods : https://docs.xarray.dev/en/stable/generated/xarray.plot.FacetGrid.html

After trying these functions with figanos and xarray.plot, I am not sure we can make them work the way we want. I only get mutiple facetgrids and they are not put on top of each other.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[juliettelavoie]

Re: 2 facetgrid together. I haven't found a way to combine facetgrid yet. But it it possible to modify individually each subplot like so:

import numpy as np
air = xr.tutorial.open_dataset("air_temperature").air
ax=air.isel(lon=10, lat=[19, 21, 22]).plot.line(x="time", col='lat')
ax.axes[0][0].plot([np.datetime64('2014-01-02T00:00:00.000000000')],[295], '.')

Hence it would be possible (though not very elegant) to add a scatter map by hand on top of a gridmap.

[sarahclaude]

Should we add arguments similar to vmin & vmax for the marker size in scattermap? If not there's no way to use the same values for multiple plots in a loop (it would only work when using 'col' or 'row' in plot_kw)

[vindelico]

as-tu réussi à faire une gridmap et scattermap ensemble sur une figure? Je pense que ça va être plus compliquer qu'on pensait.. https://stackoverflow.com/questions/44158276/combine-different-seaborn-facet-grids-into-single-plot

The link suggests you are asking about having a facetgrid with gridmap(s) and scattermap(s) alongside each other, i.e. different tiles of the facetgrid have either gridmap(s) or scattermap(s). A different way of combining them is my use case for stacking facetgrids of gridmaps and of scattermaps, so that the scattermaps are superposed over the gridmaps, in other words: dots on maps. I will look into that.

I think we should try these functions instead since xarray creates its own facetgrid object with its own methods : https://docs.xarray.dev/en/stable/generated/xarray.plot.FacetGrid.html

After trying these functions with figanos and xarray.plot, I am not sure we can make them work the way we want. I only get mutiple facetgrids and they are not put on top of each other.

I looked into xarray.plot.FacetGrid.map* methods (and the corresponding methods in seaborn.FacetGrid for pandas.DataFrame). It seems that the mapping can only use data from the underlying dataset/dataframe from a first use of row/col/col_wrap args, generating the facetgrid. Correct me if I'm wrong but I believe overlaying facetgrids would require to add another dataset as an attribute of the facetgrid object, and modifying the map* methods, particularly when the structure is different, which is so with the "dots on maps" use case and our differing data structure for grids vs station data which can not be merged into one xarray.Dataset.

That said, iterating over the facets (axs) in the facetgrid is not so un-elegant at this point.

[vindelico]

It seems like placing the title above the plot with y>1 only works in notebooks:
https://stackoverflow.com/questions/55767312/how-to-position-suptitle#comment98213058_55768955
I have successfully used it in notebooks, but it fails for me in SciView/PyCharm.

[sarahclaude]

I can remove the y=1.05, but I have noticed that playing with the figsize also helps with suptitle

[sarahclaude]

Gridmap, hatchmap and scattermap should now be working with facetgrids

[Zeitsperre]

FYI, this is the sole offending warning on ReadTheDocs:

checking consistency... /home/docs/checkouts/readthedocs.org/user_builds/figanos/checkouts/136/docs/notebooks/figanos_multiplots.ipynb: WARNING: document isn't included in any toctree

[juliettelavoie]

comme il n'y a pas d'exemple ici, je l'enleverrais ou j'écrirais "Section coming soon".

Reply via ReviewNB

[juliettelavoie]

juste une question d'esthétique. J'ajouterais        legend_kw={'ncol':4,'bbox_to_anchor':(0.15, 0.05)

Reply via ReviewNB

[juliettelavoie]

avec xclim 0.47, cette cellule brise parce que change_significance a une nouvelle signature.

https://xclim.readthedocs.io/en/stable/changes.html

change_significance va m[eme disparaitre à 0.49. but this can be fixed in another PR.

Reply via ReviewNB

[juliettelavoie]

great job!

[juliettelavoie]

@Zeitsperre is it possible that something you changed recently put a lower timeout or memory restriction on the RTD build ?

[Zeitsperre]

@juliettelavoie No, there's nothing significantly impacting to the docs that was changed from my end. No additional commands are being called, no new dependencies. The only changes I added are either commented out or they have no impact (unused locale-related changes)

[juliettelavoie]

Le probleme est dans la cellule hatchmap. C'est trop lourd de loader les 3 modèles. Je suggère de faire un exemple sans change_significance, genre juste un hatch si tas>30.

[sarahclaude]

Added facetgrids for timeseries (should be the last added function on this PR)

[juliettelavoie]

I might change this to just " Multiple Plots"

Reply via ReviewNB

[juliettelavoie]

I think it is ready to merge! so much work in here! nice job!