How to plot a Pandas bar chart from a DataFrame in Python?

In this tutorial we’ll learn the basics of charting a bar graph out of a DataFrame using Python. If you are using Pandas for data wrangling, and all you need is a simple chart you can use the basic built-in Pandas plots.

Create Bar plot from Pandas DataFrame

Proceed as following to plot a bar chart in pandas:

• Create a pandas DataFrame from a file, database or dictionary.
• Use the DataFrame plot() method to define your chart.
• Customize your chart as needed by resizing it, add a legen, set your chart tile, set your axes ticks labels, etc’.

Importing data to the Python DataFrame

We’ll start with creating a pandas DataFrame by importing data using the read_csv() method.

import pandas as pd

my_df =  pd.read_csv('interviews.csv')

print (my_df)

Here’s our DataFrame populated with some hr interview data:

	language	first_interview	second_interview
1	Kotlin	70.0	81.0
2	VisualBasic	84.0	75.0
3	PHP	82.0	91.0
4	Python	86.0	81.0

Make a Bar graphs from a DataFrame

If we want to create a simple chart we can use the df.plot() method. Note that we’ll use the kind= parameter in order to specify the chart type. Several graphs are available such as histograms, pies, area, scatter, density etc’.

my_df.plot(kind='bar' , x='language');

Here’s the early version of our very simple side by side bar chart:

Resize Pandas charts and adding a title

Modifying the Pandas chart size is easy using the figsize parameter. We are also use the title parameter to define the chart header content.

my_df.plot(kind='bar' , x='language', title='Interviews per month', figsize= (11,6));

Here’s the chart:

Making an horizontal Python graph

By using the barh chart type, we can transpose the graph rendering.

my_df.plot(kind='barh' , x='language', title='Interviews per month');

Stacked Python plot with Pandas

We can easily stack our bars chart as needed using the stacked parameter.

# stacked pandas bar graph
my_df.plot(kind='bar' , x='language', stacked=True, figsize= (11,6));

Here’s the chart:

Change the color of our Python chart

The DataFrame.plot() method also delivers capability to map the chart to a known color map (cmap). Here’s an example:

#change color maps
my_df.plot(kind='bar' , x='language', stacked=True, cmap='Dark2',figsize= (11,6));

And here’s the chart:

Modify the chart axes labels

We can also change the style of the axes labels. In this example we rotate the x-axis labels and enlarge the label font size.

my_df.plot(kind='bar' , x='language', stacked=True, cmap='Dark2', rot=30, fontsize=13, figsize= (11,6)));

And here’s the output:

Stacked columns plotting in Pandas – practical example

Step 1: Arrange your DataFrame

First step will be create a simple DataFrame containing the data that you would like to plot. In our case those would be random figures in a hiring campaign.

import pandas as pd
area = ['Java', 'R', 'Python', 'Javascript', 'Python']
applications = [150, 168, 158, 75, 98]
hired = [17,14, 17, 12, 15]
campaign = dict(area = area, hired = hired, applications  = applications )
hrdf = pd.DataFrame(data=campaign)
hrdf.head()

Let’s loo into the data:

	area	hired	applications
0	Java	17	150
1	R	14	168
2	Python	17	158
3	Javascript	12	75
4	Python	15	98

Our goal will be to create a simple stacked plot showing the number of hired vs applicants for every programming language. Looking inot the DataFrame rows we can see that we have a couple of rows (indexes 2 and 4) that pertain to the Python area.

Hence we will first group our DataFrame rows by area:

hr_grp = hrdf.groupby(['area']).sum()
print(hr_grp)

Here’s our group data:

	hired	applications
area
Java	17	150
Javascript	12	75
Python	32	256
R	14	168

Step 2: Render your bar chart with Pandas

Now, we’ll render our chart. As Pandas contains some basic matplotlib capability, we don’t need to import matplotlib or Seaborn to render simple charts.

hr_grp.plot(kind='bar', stacked = True, title= 'Hired vs applications by Area');

Here’s our chart:

Note the usage of the stacked=True parameter. Otherwise the columns will show up once next to the other.

hr_grp.plot(kind='bar', title= 'Hired vs applications by Area');

Step 3: Customize your stacked plot

Now it is time to fix up the loo and feel of our chart. We will change the chart color, re-position the legend and tide up our chart axis.

bar_chart = hr_grp.plot(kind='bar', stacked = True, cmap = 'viridis', title= 'Hired vs applications by Area') #1 

bar_chart.legend(bbox_to_anchor= (1.3, 1)) #2 
bar_chart.set_xlabel ('Programming Language', fontsize = 13); #3
bar_chart.set_ylabel ('Hired vs Applicants', fontsize = 13); #4 

Here’s our nice stacked chart:

Explanation

• #1 – setting your chart color map.
• #2 – repositioning the legend
• #3 – setting the x label content and font
• #4 – setting the y label content and font