implement `.plot()`, `.scatterplot()` and `.lisa_map()` methods for `Moran_Local` object

[slumnitz]

implement visualisation methods for esda.moran.Moran_Local called from splot

write documentation and tests for methods

[slumnitz]

Travis depends on this pull request to be able to install splot and call splot.esda from the master branch of the splot repository.

The last error in Travis seems unrelated: ERROR: test_by_col (esda.tests.test_moran.Moran_Rate_Tester)

@sjsrey @darribas @ljwolf

[sjsrey]

@slumnitz I'm hitting an error trying to test this off of the branch slumnitz:splot-method: (All in a notebook but kludgy, so cut and paste here)

import matplotlib.pyplot as plt
import libpysal.api as lp
from libpysal import examples
from esda.moran import Moran_Local
import geopandas as gpd
link = examples.get_path('columbus.shp')
gdf = gpd.read_file(link)
y = gdf['HOVAL'].values
w = lp.Queen.from_dataframe(gdf)
w.transform = 'r'
mloc = Moran_Local(y, w)
fig, _ = mloc.plot(gdf, 'HOVAL')
plt.close(fig)

Is resulting in:

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-3-bab1aa5c2c4b> in <module>()
     10 w.transform = 'r'
     11 mloc = Moran_Local(y, w)
---> 12 fig, _ = mloc.plot(gdf, 'HOVAL')
     13 plt.close(fig)

~/Dropbox/p/pysal/src/subpackages/esda/esda/moran.py in plot(self, gdf, attribute, p, region_column, mask, mask_color, quadrant, legend, scheme, cmap, figsize)
   1008                                                         scheme=scheme,
   1009                                                         cmap=cmap,
-> 1010                                                         figsize=figsize)
   1011         return fig, axs
   1012 

~/Dropbox/p/pysal/src/subpackages/splot/splot/_viz_esda_mpl.py in plot_local_autocorrelation(moran_loc, gdf, attribute, p, region_column, mask, mask_color, quadrant, legend, scheme, cmap, figsize)
    270     lisa_cluster(moran_loc, gdf, p=p, ax=axs[1], legend=legend,
    271                  legend_kwds={'loc': 'upper left',
--> 272                  'bbox_to_anchor': (0.92, 1.05)})
    273     axs[1].set_aspect('equal')
    274 

~/Dropbox/p/pysal/src/subpackages/splot/splot/_viz_esda_mpl.py in lisa_cluster(moran_loc, gdf, p, ax, legend, legend_kwds, **kwargs)
    187                               k=2, cmap=hmap, linewidth=0.1, ax=ax,
    188                               edgecolor='white', legend=legend,
--> 189                               legend_kwds=legend_kwds, **kwargs)
    190     ax.set_axis_off()
    191     ax.set_aspect('equal')

~/anaconda3/envs/libpysal3/lib/python3.6/site-packages/geopandas/geodataframe.py in plot(self, *args, **kwargs)
    467     def plot(self, *args, **kwargs):
    468 
--> 469         return plot_dataframe(self, *args, **kwargs)
    470 
    471     plot.__doc__ = plot_dataframe.__doc__

~/anaconda3/envs/libpysal3/lib/python3.6/site-packages/geopandas/plotting.py in plot_dataframe(df, column, cmap, color, ax, categorical, legend, scheme, k, vmin, vmax, figsize, **style_kwds)
    440     if not polys.empty:
    441         plot_polygon_collection(ax, polys, values[poly_idx],
--> 442                                 vmin=mn, vmax=mx, cmap=cmap, **style_kwds)
    443 
    444     # plot all LineStrings and MultiLineString components in same collection

~/anaconda3/envs/libpysal3/lib/python3.6/site-packages/geopandas/plotting.py in plot_polygon_collection(ax, geoms, values, color, cmap, vmin, vmax, **kwargs)
     95 
     96     collection = PatchCollection([PolygonPatch(poly) for poly in geoms],
---> 97                                  **kwargs)
     98 
     99     if values is not None:

~/anaconda3/envs/libpysal3/lib/python3.6/site-packages/matplotlib/collections.py in __init__(self, patches, match_original, **kwargs)
   1731             kwargs['antialiaseds'] = [p.get_antialiased() for p in patches]
   1732 
-> 1733         Collection.__init__(self, **kwargs)
   1734 
   1735         self.set_paths(patches)

~/anaconda3/envs/libpysal3/lib/python3.6/site-packages/matplotlib/collections.py in __init__(self, edgecolors, facecolors, linewidths, linestyles, capstyle, joinstyle, antialiaseds, offsets, transOffset, norm, cmap, pickradius, hatch, urls, offset_position, zorder, **kwargs)
    168 
    169         self._path_effects = None
--> 170         self.update(kwargs)
    171         self._paths = None
    172 

~/anaconda3/envs/libpysal3/lib/python3.6/site-packages/matplotlib/artist.py in update(self, props)
    886         try:
    887             ret = [_update_property(self, k, v)
--> 888                    for k, v in props.items()]
    889         finally:
    890             self.eventson = store

~/anaconda3/envs/libpysal3/lib/python3.6/site-packages/matplotlib/artist.py in <listcomp>(.0)
    886         try:
    887             ret = [_update_property(self, k, v)
--> 888                    for k, v in props.items()]
    889         finally:
    890             self.eventson = store

~/anaconda3/envs/libpysal3/lib/python3.6/site-packages/matplotlib/artist.py in _update_property(self, k, v)
    879                 func = getattr(self, 'set_' + k, None)
    880                 if not callable(func):
--> 881                     raise AttributeError('Unknown property %s' % k)
    882                 return func(v)
    883 

AttributeError: Unknown property legend_kwds

[slumnitz]

legend_kwds is currently only available in the geopandas master branch. It has not been included in the gdf.plot() method in the last release of geopandas. Maybe @jorisvandenbossche has some info about if this is going to be included in a next release? @sjsrey I can add a warning here that the plot methods depend on the geopandas master branch?

[ljwolf]

@slumnitz sorry for the delay in reviewing. I believe the issue here is due to the way the testing methods for unittest have not kept up to date with numpy's revisions.

Can you try changing self.assertEquals to numpy.testing.assert_allclose where that fails in the by_col test? I believe that would resolve.

[slumnitz]

@ljwolf I tried adapt your suggestion. It still failed for me setting atol=ATOL and rtol=RTOL. Then I tried testing with values atol=0, rtol=1e-7 and so on, but it still kept failing. I am not sure what is happening but it seems like the actual p-value might be off?

[ljwolf]

you're right. Note that a seed is not set in this testing class's setUp.

We need to set the seed and ensure that the resulting pval is right in the comparison.

on my box I get anything from .01 to .011.

[slumnitz]

I fixed the by_col issue and found the source of the other problem: https://github.com/pysal/splot/pull/16

Once this is merged Travis will hopefully pass.

[slumnitz]

Another issue with dependencies:

ModuleNotFoundError: No module named 'seaborn'

This should be installed when splot is installed. I will try to fix this in splot first. I'll include that in https://github.com/pysal/splot/pull/19

[sjsrey]

@slumnitz can we close this?

[slumnitz]

@sjsrey this PR was opened to call splot plotting methods from esda. I recall it was not merged for a while, because of open PRs in splot. If it is still of interest to call splot methods with .plot() in esda, I am happy to clean this up. I am not sure how difficult this will be since no-one has found out why splot fails at the moment. Otherwise it can be closed or closed and added as an issue in splot for future implementation? Your choice. :)

[jGaboardi]

@slumnitz I left notes for you regarding the splot testing failures. Although I have not had the time to track down the specific trigger for the segfault, the issue seems to be memory (code 139).