Pandas Pivot Table | Create Pivot Table Using Pandas in Python

Pivot tables – the Swiss Army Knife of data analysis

I love how quickly I can analyze data using pivot tables. With one click of my mouse, I can drill down into the granular details about a certain product category or zoom out and get a high-level overview of the data at hand. Microsoft Excel users will be intimately familiar with these pivot tables. They’re the most used feature of Excel, and it’s easy to see why! But did you know that you can build these pivot tables using Pandas in Python? A pandas pivot table is easy to create and works just the same.

That’s right! The wonderful Pandas library offers a list of functions, among which a function called pivot_table is used to summarize a feature’s values in a neat two-dimensional table. The pivot table is similar to the dataframe.groupby() function in Pandas. We’ll see how to build such a pivot table in Python here. We can even use Pandas pivot table along with the plotting libraries to create different visualizations. Pivot tables offer a ton of flexibility for me as a data scientist. I’ll be honest – I rely on them a lot during the exploratory data analysis phase of a data science project.

Learning Objectives

• Learn to create a python pivot table using Pandas.
• Learn to use different aggregation functions on the features of the created pivot table in pandas.
• Learn to handle missing data in the pandas pivot table.

Sample Dataset for Testing Pandas Pivot Table

I’m sure you’ve come across the Titanic dataset in your data science journey. It’s among the first datasets we pick up when we’re ready to explore a project. I’ll be using that to show you the efficacy of the pivot_table function.

Let’s import the relevant libraries:

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
plt.style.use('ggplot')

For all those who forgot what the Titanic dataset looks like, I present to you the dataset!

df = pd.read_csv('drive/My Drive/AV/train.csv')
df.head()Table of Contents
Sample Dataset for Testing Pandas Pivot Table
Building a Pivot Table Using Pandas
Conclusion
Frequently Asked Questions
Q1. How do you create a pivot table in Python?
Q2. What is the DataFrame.pivot method?
Q3. what is the difference between the pivot and pivot_table methods in Python Pandas?
Sample Dataset for Testing Pandas Pivot Table
I’m sure you’ve come across the Titanic dataset in your data science journey. It’s among the first datasets we pick up when we’re ready to explore a project. I’ll be using that to show you the efficacy of the pivot_table function.

Let’s import the relevant libraries:

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
plt.style.use('ggplot')
For all those who forgot what the Titanic dataset looks like, I present to you the dataset!
df = pd.read_csv('drive/My Drive/AV/train.csv')
df.head()

I will be dropping multiple columns to make it easier to analyze the data and demonstrate the capabilities of the pivot_table function:

df.drop(['PassengerId','Ticket','Name'],inplace=True,axis=1)

The above syntax drops the columns with column names, 'PassengerId', 'Ticket', and 'Name', from the dataset.

Building a Pivot Table Using Pandas

Time to build a pivot table in Python using the awesome Pandas library! We will explore the different facets of a python pivot table in pandas in this article and build an awesome, flexible excel-style pivot table from scratch.

How to Group Data Using Index in a Pivot Table?

• pivot_table requires data, and an index parameter
• data is the Pandas dataframe you pass to the function.
• index is the feature that allows you to group your data. The index feature will appear as an index in the resultant table. Generally, categorical columns are used as indexes.

I will be using the ‘Sex’ column as the index for now:

#a single index
table = pd.pivot_table(data=df,index=['Sex'])
table

We can instantly compare all the feature values for both the genders. Now, let’s visualize the finding.

Python Code:

Well, the female passengers paid remarkably more for the tickets than the male passengers.

You can learn more about how to visualize your data here.

How to Run a Pivot With a Multi-Index?

You can even use more than one feature as an index to group your data. This increases the level of granularity in the resultant table, and you can get more specific with your findings:

#multiple indexes
table = pd.pivot_table(df,index=['Sex','Pclass'])
table

Using multiple indexes on the dataset enables us to concur that the disparity in ticket fare for female and male passengers was valid across every Pclass on Titanic.

Different Aggregation Functions for Different Features

The values shown in the table are the result of the summarization that aggfunc applies to the feature data. aggfunc is an aggregate function that pivot_table applies to your grouped data.

By default, it is np.mean(), but you can use different aggregate functions for different features too! Just provide a dictionary as an input to the aggfunc parameter with the feature name as the key and the corresponding aggregate function as the value.

I will be using np.mean() to calculate the mean for the ‘Age’ feature and np.sum() to calculate the total survivors for the ‘Survived’ feature:

#different aggregate functions
table = pd.pivot_table(df,index=['Sex','Pclass'],aggfunc={'Age':np.mean,'Survived':np.sum})
table

The resultant table makes more sense in using different aggregating functions for different features.

Aggregate on Specific Features With Values Parameter

But what are you aggregating on? You can tell Pandas the feature(s) to apply the aggregate function on in the value parameter. The value parameter is where you tell the function which features to aggregate on. It is an optional field, and if you don’t specify this value, then the function will aggregate all the numerical features of the dataset:

table = pd.pivot_table(df,index=['Sex','Pclass'],values=['Survived'], aggfunc=np.mean)
table

table.plot(kind='bar');

The survival rate of passengers aboard the Titanic decreased with a degrading Pclass among both the genders. Moreover, the survival rate of male passengers was lower than that of female passengers in any given Pclass.

Find the Relationship Between Features With Columns Parameter

Using multiple features as indexes is fine, but using some features as columns will help you to intuitively understand the relationship between them. Also, the resultant table can always be better viewed by incorporating the columns parameter of the pivot_table.

The columns parameter is optional and displays the values horizontally on the top of the resultant table. Both columns and the index parameters are optional, but using them effectively will help you to intuitively understand the relationship between the features.

#columns
table = pd.pivot_table(df,index=['Sex'],columns=['Pclass'],values=['Survived'],aggfunc=np.sum)
table

Using Pclass as a column is easier to understand than using it as an index:

table.plot(kind='bar');

Handling Missing Data

pivot_table even allows you to deal with the missing values through the parameters dropna and fill_value:

• dropna allows you to drop the null values in the grouped table whose all values are null
• fill_value parameter can be used to replace the NaN values in the grouped table with the values that you provide here

#display null values
table = pd.pivot_table(df,index=['Sex','Survived','Pclass'],columns=['Embarked'],values=['Age'],aggfunc=np.mean)
table

I will be replacing the NaN values with the mean value from the ‘Age’ column:

#handling null values
table = pd.pivot_table(df,index=['Sex','Survived','Pclass'],columns=['Embarked'],values=['Age'],aggfunc=np.mean,fill_value=np.mean(df['Age']))
table

Conclusion

In this article, we explored the different parameters of the awesome pivot_table function and how it allows you to easily summaries the features in your dataset through a single line of code. If you are new to Python programming and want to learn more about data analysis with Python, I highly recommend you explore our free Python for Data Science and Pandas for Data Analysis in Python courses.

Key Takeaways

• We can create python pivot table in pandas using the pivot_table function.
• We can pass different parameters in the pivot_table function to aggregate data, handle missing values, etc.

Frequently Asked Questions

Aniruddha Bhandari

I am on a journey to becoming a data scientist. I love to unravel trends in data, visualize it and predict the future with ML algorithms! But the most satisfying part of this journey is sharing my learnings, from the challenges that I face, with the community to make the world a better place!