In [1]:
name = '2015-11-27-meeting-summary'
title = 'Subplots in matplotlib'
author = 'Denis Sergeev'
tags = 'matplotlib, jupyter, binder'
In [2]:
from nb_tools import connect_notebook_to_post
from IPython.core.display import HTML
html = connect_notebook_to_post(name, title, tags, author)
Today we discussed some of the basic matplotlib functions and also had a look at different ways of running Jupyter Notebooks.
In [3]:
%matplotlib inline
In [4]:
import matplotlib.pyplot as plt
import numpy as np
To show figure area:
In [5]:
plt.rcParams['figure.facecolor'] = '0.8' # grey background
Sample 1D array of size 100:
In [6]:
x = np.random.rand(100)
Two simplest ways of creating a bunch of subplots within one figure are shown below.
In [7]:
fig = plt.figure(figsize=(5,5))
ax_a = fig.add_subplot(131)
ax_b = fig.add_subplot(132)
ax_c = fig.add_subplot(133)
for iax in [ax_a, ax_b, ax_c]:
iax.plot(x)
In [8]:
fig, (ax_a, ax_b, ax_c) = plt.subplots(ncols=3)
for iax in [ax_a, ax_b, ax_c]:
iax.plot(x)
Using the same handy subplots() command, we can create an $2\times 3$ grid of subplots.
In [9]:
fig, axs = plt.subplots(nrows=3, ncols=2)
for iax in axs.flat:
iax.plot(x, color='r')
If the subplots have same axes, it's natural to leave axis labels only in outermost subplots. We can do this setting sharex and sharey keywords in subplots() function to True.
In [10]:
fig, axs = plt.subplots(nrows=3, ncols=2, sharex=True, sharey=True)
for iax in axs.flat:
iax.plot(x, color='r')
A more sophisticated grid of subplots can be created using matplotlib.gridspec submodule. The quickstart guide can be found on this page:
The issue of getting rid of white space in figures was also mentioned during the meeting. Matplotlib has the feature called tight_layout() that in some cases is enough to fit all subplots in a figure.
In [11]:
import cartopy
import cartopy.crs as ccrs
In [14]:
fig = plt.figure(figsize=(30, 10))
ax = fig.add_subplot(111, projection=ccrs.PlateCarree())
ax.coastlines()
gl = ax.gridlines(draw_labels=True)
gl.xlabels_top = gl.ylabels_right = False
gl.xformatter = cartopy.mpl.gridliner.LONGITUDE_FORMATTER
gl.yformatter = cartopy.mpl.gridliner.LATITUDE_FORMATTER
lon, lat = np.linspace(-100,50,10), np.linspace(10,60,20)
arr = np.random.rand(20,10)
c = ax.contourf(lon, lat, arr)
cb = plt.colorbar(c, ax=ax, orientation='horizontal',
fraction=0.046, pad=0.04) # magically sets colorbar size correctly...
fig.tight_layout(pad=0)
And by the way, ax is not just a matplotlib.axes._subplots.AxesSubplot instance anymore, but has geoaxes attributes:
In [15]:
type(ax)
Out[15]:
Jupyter Notebooks can be launched not only on a local PC, but also on a remote host with the help of some web-services.
We took a glimpse at
In [16]:
HTML(html)
Out[16]: