How to make tabular legend using matplotlib and python

Question

I am plotting a choropleth map in python from a shapefile and i want to customize the legend of the plot, i am using the code bellow:

import pandas as pd
import pysal as ps
import geopandas as gp
import numpy as np
import matplotlib.pyplot as plt

pth = 'outcom.shp'
tracts = gp.GeoDataFrame.from_file(pth)
ax = plot_dataframe(tracts, column='Density', scheme='QUANTILES', k=4, colormap=plt.cm.Blues, legend=True)
plt.show()

Besides, i am using a small patch that i found here http://nbviewer.ipython.org/gist/jorisvandenbossche/d4e6efedfa1e4e91ab65 in order to visualize the legend.

here's my result : But, i need something similar to this :

so my question now, is how can i have a customized legend

Answer 1

You may use a plt.table as a legend.

import matplotlib.pyplot as plt
import numpy as np

valeur = np.array([.1,.45,.7])
text=[["Faible","Ng<1,5" ],["Moyenne","1,5<Ng<2,5"],[u"Elevée", "Ng>2,5"]]
colLabels = ["Exposition", u"Densité"]


tab=plt.table(cellText=text, colLabels=colLabels, 
                    colWidths = [0.2,0.2], loc='lower right', 
                    cellColours=plt.cm.hot_r(np.c_[valeur,valeur]))

plt.show()

In order to link this table to a contourf plot, you may do as follows:

from matplotlib import pyplot as plt
import numpy as np

a = np.sort(np.random.rand(100)).reshape(10,10)*4

levels = np.array([0,1.5,2.5,4])
sm = plt.contourf(a, levels = levels, cmap=plt.cm.hot_r )


text=[["Faible","Ng<1,5" ],["Moyenne","1,5<Ng<2,5"],[u"Elevée", "Ng>2,5"]]
colLabels = ["Exposition", u"Densité"]

col = levels[:-1] + np.diff(levels)/2.
cellcol = sm.cmap(sm.norm(np.c_[col,col]))

tax = plt.gcf().add_axes([0,0,1,1])
tab=tax.table(cellText=text, colLabels=colLabels, 
                    colWidths = [0.2,0.2], loc='lower left', 
                    cellColours=cellcol )

tax.axis("off")
plt.show()