Get information from plt.hexbin

I have been using hexbing to try to improve the SPCF method, so for that I would need to get information from the plot to modify the data and then replot it with the new values.

The most important data I need to modify to replot is the number of points per cell and the C value of that grid cell.

Any way I can get such data?

Answer

matplotlib.pyplot.hexbin returns a PolyCollection object with methods get_offsets and get_array which provide exactly what you’re looking for. From the hexbin docs:

Returns: PolyCollection
A PolyCollection defining the hexagonal bins.

• PolyCollection.get_offsets contains a Mx2 array containing the x, y positions of the M hexagon centers.
• PolyCollection.get_array contains the values of the M hexagons.

If marginals is True, horizontal bar and vertical bar (both PolyCollections) will be attached to the return collection as attributes hbar and vbar.

Example

As an example, I’ll create a simple hexbin plot and grab the PolyCollection object returned by plt.hexbin:

In [2]: x = np.linspace(1, 20, 500)

In [3]: y = np.log(x) + np.random.random(size=500) - 0.5

In [4]: coll = plt.hexbin(x, y, gridsize=20)

You can explore the collection object and grab the arrays returned by get_offsets() and get_array():

In [23]: type(coll)
Out[23]: matplotlib.collections.PolyCollection

In [24]: offsets = coll.get_offsets()

In [25]: offsets
Out[25]:
array([[ 9.99999981e-01, -3.48964611e-01],
       [ 9.99999981e-01, -3.91102264e-03],
       [ 9.99999981e-01,  3.41142565e-01],
       ...
       [ 1.95250000e+01,  2.58399089e+00],
       [ 1.95250000e+01,  2.92904448e+00],
       [ 1.95250000e+01,  3.27409806e+00]])

In [26]: arr = coll.get_array()

In [27]: arr
Out[27]:
masked_array(data=[ 2.,  4.,  0.,  ...,  7.,  2.,  4.],
             mask=False,
       fill_value=1e+20)

In [28]: offsets.shape
Out[28]: (472, 2)

In [29]: arr.shape
Out[29]: (472,)

You could manipulate these arrays however you want, and then plot them again using the x, y, and C count argument to hexbin. If you don’t change the values of offsets and arr, this plots the exact same image:

In [30]: plt.hexbin(offsets[:, 0], offsets[:, 1], C=arr, gridsize=20)