pyplotthemes

A wrapper around matplotlib with themes.

Usage:

Instead of doing the typical

from matplotlib import pyplot as plt

you instead import the theme you want

from pyplotthemes import classictheme as plt

Every method/attribute in pyplot is available in the themes so this can be a one line replacement to get away from the default settings.

Why another theme package?

There are several packages available that try to change the way your plots look. Most change the matplotlib rcParams, but go no further. In addition, they assume you have not edited your rcParams file beforehand, and it's not trivial to use many different themes in the same program.

This package tries to achieve the following:

Every plotting call assumes nothing about styles, and will reset the rcParams before every call. This makes it possible to use several themes in one program and ensures consistency even if you've edited your rcParams file.
Tweak things that can't be tweaked with rcParams, like axes positioning, legend alpha, etc.
Allow everything to be customized. You can override all rcParam values in the theme constructor or before/after plotting calls with the setstyle method.

Examples



In [1]:

    
%matplotlib inline
import numpy as np

# Here I import all themes, you might want to do "from pyplotthemes import theme as plt" instead
from pyplotthemes import basetheme, classictheme, prettytheme


# For easy iteration in the examples
themes = [basetheme, classictheme, prettytheme]
names = [theme.__class__.__name__ for theme in themes]
# Set Latex to true in non-base themes
for name, theme in zip(names, themes):
    theme.latex = (name != basetheme.__class__.__name__)

Basic plot with legend

Note the differences in backgrounds, frames, and ticks.



In [2]:

    
def plottest(theme):
    # Set the random seed for consistency
    np.random.seed(0)

    theme.figure(figsize=(5,4))

    # Show the whole color range
    for i in range(8):
        y = np.random.normal(size=1000).cumsum()
        x = np.arange(1000)

        theme.plot(x, y, label=str(i))

    theme.legend(loc='best', ncol=3)
    

for name, theme in zip(names, themes):
    plottest(theme)
    theme.title(name)

Polar axes

To get a polar projection you need to create the axes before plotting. Either with axes or with subplot.



In [3]:

    
def plottest(theme):
    angle = np.linspace(0, 3*2*np.pi, 100)
    r = np.linspace(0, 1, 100)

    theme.figure(figsize=(5,4))
    # Create a polar axes. Either polar=True, or projection='polar'
    theme.axes(projection='polar')
    # Then just call plot like normal
    theme.plot(angle, r)
    

for name, theme in zip(names, themes):
    plottest(theme)
    theme.title(name)

Logarithmic axes

Commonly used when the range of values are very large.



In [11]:

    
def logtest(theme, name):
    theme.figure(figsize=(15,4))

    for sub in range(3):
        theme.subplot(131 + sub)
        theme.title(name)
        # Show the whole color range
        for i in range(8):
            x = np.linspace(0, 1, 10)
            y = np.exp(-x*(i + 1))

            if sub == 0:
                theme.semilogx(x, y, label=str(i), marker='o')
            elif sub == 1:
                theme.semilogy(x, y, label=str(i), marker='o')
            else:
                theme.loglog(x, y, label=str(i), marker='o')
                
            theme.legend(loc='best', ncol=3)
    

for name, theme in zip(names, themes):
    logtest(theme, name)

Histograms

Notice that some non-base themes have visible grids which makes the actual values easier to read.



In [5]:

    
def histtest(theme):
    np.random.seed(0)
    theme.figure(figsize=(5,4))
    
    x = np.random.normal(size=100)
    
    theme.hist(x, label='A label')
    theme.legend(loc='best')
    

for name, theme in zip(names, themes):
    histtest(theme)
    theme.title(name)

Boxplot



In [6]:

    
def boxtest(theme):
    np.random.seed(0)
    theme.figure(figsize=(5,4))
    
    x = []
    for j in range(5):
        x.append(np.random.exponential(3*(j + 1), size=100))
    
    theme.boxplot(x)
    #theme.legend(loc='best')
    

for (name, theme) in zip(names, themes):
    boxtest(theme)
    theme.title(name)

Heatmaps



In [8]:

    
def heattest(theme, mins):
    np.random.seed(10)
    x = np.random.normal(size=(10, 10))
    
    if mins < 0:
        # Force into negative
        x = x - np.max(x)
    elif mins > 0:
        # Force into positive
        x = x - np.min(x)
    
    p = theme.pcolormesh(x)
    theme.colorbar(p)
    
    

for (name, theme) in zip(names, themes):
    theme.figure(figsize=(15,4))
    for mins, a in zip([-1, 0, 1], [131, 132, 133]):
        theme.subplot(a)
        heattest(theme, mins)
        
        t = name
        if mins < 0:
            t += " only negative"
        elif mins == 0:
            t += " both +/-"
        else:
            t += " only positive"
        theme.title(t)