Pandas Plotting
last modified November 17, 2026 Python 3.15
Pandas can turn a DataFrame or Series into a chart with a single method call.
The plot method is a thin convenience layer built on top of
Matplotlib, the underlying plotting
library; it passes the data to Matplotlib and returns the axes so they can be
customised further. If you want full control of a figure, the
Matplotlib line chart tutorial and the
rest of that section cover the same machinery directly.
The scripts below use the non-interactive Agg backend, which is
set before pyplot is imported so the code runs without a display. Each script
saves its figure to a PNG file on disk and prints a short confirmation; the
output blocks therefore contain printed text rather than the images themselves.
A line plot
A line plot connects data points in order and is the usual choice for trends
over time or another ordered variable. Here the x and
y arguments pick the columns to draw, while
title, xlabel and ylabel add text and
figsize sets the figure's width and height in inches.
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import pandas as pd
df = pd.DataFrame({'month': [1, 2, 3, 4], 'sales': [12, 18, 15, 24]})
ax = df.plot(kind='line', x='month', y='sales', title='Monthly sales',
xlabel='Month', ylabel='Sales', figsize=(6, 4))
plt.gcf().savefig('/tmp/opencode/agentH_plot_line.png')
print('Saved /tmp/opencode/agentH_plot_line.png')
The chart is drawn in memory and then written to the PNG file with
savefig; the printed line simply confirms that the file was
created. Calling matplotlib.use('Agg') before importing
pyplot matters: once pyplot is imported, the backend is fixed.
$ python plot_line.py Saved /tmp/opencode/agentH_plot_line.png
A bar plot
A bar plot compares categories by drawing one rectangle per category, with the
height showing the value. It is appropriate when the labels have no natural
order. The same labelling arguments work here, and the categories are taken
from the column named by x.
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import pandas as pd
df = pd.DataFrame({'city': ['A', 'B', 'C', 'D'], 'population': [5, 8, 3, 6]})
ax = df.plot(kind='bar', x='city', y='population', title='Population by city',
xlabel='City', ylabel='Population (millions)', figsize=(6, 4))
plt.gcf().savefig('/tmp/opencode/agentH_plot_bar.png')
print('Saved /tmp/opencode/agentH_plot_bar.png')
Four bars are drawn, one for each city, and the result is saved alongside the
first image. The ax variable holds the Matplotlib axes object, so
further Matplotlib calls can adjust ticks, colours or legends after the plot
has been created.
$ python plot_bar.py Saved /tmp/opencode/agentH_plot_bar.png
A histogram
A histogram shows the distribution of a single numeric column: the
values are sorted into ranges called bins, and each bar shows how many values
fell into that bin. It differs from a bar plot, where each bar represents a
named category rather than a range of numbers. The bins argument
controls how many ranges are used.
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import pandas as pd
values = pd.Series([1, 2, 2, 3, 3, 3, 4, 4, 5])
ax = values.plot(kind='hist', bins=5, title='Value distribution',
xlabel='Value', ylabel='Frequency', figsize=(6, 4))
plt.gcf().savefig('/tmp/opencode/agentH_plot_hist.png')
print('Saved /tmp/opencode/agentH_plot_hist.png')
The Series has nine values and five bins, so the bars show how those values are spread from 1 to 5. Histograms are called directly on a Series here, but DataFrames support them as well.
$ python plot_hist.py Saved /tmp/opencode/agentH_plot_hist.png
A box plot
A box plot summarises a distribution with its median, quartiles and whiskers, and marks values that lie far from the rest as outliers. Drawing one box per column makes it easy to compare the spread of several numeric columns at a glance.
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import pandas as pd
df = pd.DataFrame({'A': [10, 12, 11, 13, 12], 'B': [20, 18, 22, 19, 21]})
ax = df.plot(kind='box', title='Box plot of A and B',
ylabel='Value', figsize=(6, 4))
plt.gcf().savefig('/tmp/opencode/agentH_plot_box.png')
print('Saved /tmp/opencode/agentH_plot_box.png')
Two boxes appear, one for column A and one for
B. There is no meaningful horizontal axis here, so only the
vertical axis is labelled.
$ python plot_box.py Saved /tmp/opencode/agentH_plot_box.png
A scatter plot
A scatter plot places one dot at the intersection of two numeric values and is
used to look for a relationship between them. Unlike the line plot, no
connection is drawn between the points. Both coordinates come from columns
chosen with x and y.
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import pandas as pd
df = pd.DataFrame({'height': [150, 160, 170, 180, 190],
'weight': [50, 55, 65, 75, 85]})
ax = df.plot(kind='scatter', x='height', y='weight', title='Height vs weight',
xlabel='Height (cm)', ylabel='Weight (kg)', figsize=(6, 4))
plt.gcf().savefig('/tmp/opencode/agentH_plot_scatter.png')
print('Saved /tmp/opencode/agentH_plot_scatter.png')
Five points are drawn, one per row of the DataFrame. The upward pattern shows that the two columns rise together in this small sample; the plot itself simply places the pairs on the axes.
$ python plot_scatter.py Saved /tmp/opencode/agentH_plot_scatter.png
Multiple subplots
A subplot is one of several separate axes drawn inside a single
figure. Passing subplots=True gives each column its own panel,
which is useful for comparing series with different scales. A list of titles
labels the panels, figsize makes the figure tall enough for both,
and tight_layout keeps the labels from overlapping.
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import pandas as pd
df = pd.DataFrame({'sales': [12, 18, 15, 24], 'costs': [8, 9, 10, 11]})
axes = df.plot(subplots=True, kind='line', figsize=(6, 6),
title=['Sales', 'Costs'])
fig = plt.gcf()
fig.tight_layout()
fig.savefig('/tmp/opencode/agentH_plot_subplots.png')
print('Saved /tmp/opencode/agentH_plot_subplots.png')
The two panels are stacked vertically, one titled Sales and the
other Costs, and saved as a single image. The variable
axes holds the collection of axes; here we fetch the current
figure with plt.gcf() to adjust and save the whole image at once.
$ python plot_subplots.py Saved /tmp/opencode/agentH_plot_subplots.png
A stacked area chart
An area chart fills the space under each line, and a stacked area chart piles
the series on top of one another so the total is visible as well as the parts.
Pandas exposes this through the dedicated plot.area accessor,
which is a shortcut for plot(kind='area').
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import pandas as pd
df = pd.DataFrame({'A': [1, 2, 3, 4], 'B': [4, 3, 2, 1]})
ax = df.plot.area(title='Stacked area chart', xlabel='Index', ylabel='Value',
figsize=(6, 4))
plt.gcf().savefig('/tmp/opencode/agentH_plot_area.png')
print('Saved /tmp/opencode/agentH_plot_area.png')
The two filled regions add up to five at every point, so the top edge is flat while the boundary between the regions moves. The figure is saved to disk and the confirmation is printed, just as in the earlier examples.
$ python plot_area.py Saved /tmp/opencode/agentH_plot_area.png
Source
Pandas DataFrame.plot Documentation, Pandas Visualization Documentation, and Matplotlib savefig Documentation
In this article we have drawn line, bar, histogram, box, scatter and stacked
area charts with Pandas' plot method, arranged panels with
subplots, and labelled figures with title,
xlabel, ylabel and figsize. For the
underlying library see the Matplotlib
tutorial, and for preparing data before plotting see
Pandas grouping data.
Author
List all Pandas tutorials.