19

I really like pandas to handle and analyze big datasets. So far, I have mostly used matplotlib for plotting but now want to use pandas own plot functionalities (based on matplotlib) since it needs less code and seems to be sufficient for me in most cases. Especially the subplots to have a guick glance at big dataframes like in the following example..

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

# Generate random data
df = pd.DataFrame(np.random.randn(96,12),
                  columns=['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J',
                           'K', 'L'])

# Plotting
df.plot(kind='line', subplots=True, grid=True, title="Sample Data (Unit)",
        layout=(4, 3), sharex=True, sharey=False, legend=True,    
        style=['r', 'r', 'r', 'g', 'g', 'g', 'b', 'b', 'b', 'r', 'r', 'r'],
        xticks=np.arange(0, len(df), 16))

enter image description here

..which brings me to my questions:

1.) How can I place all legends in the subplots at the same place (e. g. centered, outside, topright)?

2.) Can I somehow use matplotlibs "Tight Layout" (http://matplotlib.org/users/tight_layout_guide.html) for the plot?

Thanks in advance!

Community
  • 1
  • 1
Cord Kaldemeyer
  • 6,405
  • 8
  • 51
  • 81

1 Answers1

25

  1. You can have all the legends in the same place, but you would have to create them in a separate step.

    # Plotting
df.plot(kind='line', subplots=True, grid=True, title="Sample Data (Unit)",
    layout=(4, 3), sharex=True, sharey=False, legend=False,    
    style=['r', 'r', 'r', 'g', 'g', 'g', 'b', 'b', 'b', 'r', 'r', 'r'],
    xticks=np.arange(0, len(df), 16))

for ax in plt.gcf().axes:
    ax.legend(loc=1)

  2. Sure. just use plt.tight_layout() before you show or savefig. Compare the two examples below created with and without tight_layout.

Without tight_layout():

enter image description here

With tight_layout():

enter image description here

tmdavison
  • 64,360
  • 12
  • 187
  • 165
  • @tom I was looking for help with the tight_layout() function, which was not working for me when using pandas and the subplot feature. However, I found a way using subplots_adjust(). Thanks for your attention – Fourier Oct 28 '16 at 13:38