[GIS] Adding custom names to a GeoPandas legend

Question

I have a shapefile which has an attribute table with a column I would like to make a map/plot of. The attribute values are numerical (integer). I have made two dicts to map the colors and names I want to these integers.

Palette = {0: 'black',
       20: '#FFBB22',
       30: '#FFFF4C',
       40: '#F096FF',
       80: '#0032C8',
       90: '#0096A0',
       112: '#009900',
       114: '#00CC00',
       116: '#007800',
       124: '#A0DC00',
       126:'#648C00'}

names  = {0: 'NAN',
       20: 'Shrubs',
       30: 'Herbaceous',
       40: 'Cultivated',
       80: 'Permanent Water',
       90: 'Herbaceous Wetland',
       112: 'Closed Forest: Evergreen',
       114: 'Closed Forest: Deciduous broad leaf',
       116: 'Closed forest: Other',
       124: 'Open forest: Deciduous broad leaf',
       126:'Open forest: Other'}

However, while I can map the colors to the right values, I cannot get the legend to show the right names. The legend comes up empty and I get a message that No handles with labels found to put in legend.

My code is:

fig, ax = plt.subplots(figsize=(5, 5))

# Loop through each attribute value and assign each
# with the correct color & width specified in the dictionary
for ctype, data in map_df.groupby('landcovermode'):
    color = Palette[ctype]
    label = names[ctype]
    data.plot(color=color,
      ax=ax,
      label=label,legend=True)


# Place legend in the lower right hand corner of the plot
ax.legend(loc='lower right',
  fontsize=15,
  frameon=True)

ax.set_axis_off()

How do I get the legend to read my labels from the dict?

Answer 1

As discussed here, matplotlib doesn't support the automatic creation of legend handlers for polygons. Your code works fine when using Point or LineString geometries. The solution would be to create a custom legend handle (ref), which could be done the following way:

from matplotlib.patches import Patch

fig, ax = plt.subplots(figsize=(5,5))
pmarks = []
for ctype, data in map_df.groupby('landcovermode'):
    color = Palette[ctype]
    label = names[ctype]
    data.plot(ax=ax, color=color)
    pmarks.append(Patch(facecolor=color, label=label))

handles, _ = ax.get_legend_handles_labels()
ax.legend(handles=[*handles,*pmarks], loc='lower right')

ax.set_axis_off()

This will be the output using some random test data: