Pandas: What is a MultiIndex and how to create one

Updated: February 20, 2024 By: Guest Contributor Post a comment


In the world of data analysis and manipulation, Pandas stands out as one of the most powerful and versatile libraries in Python. A significant feature that enhances Pandas’ capability to handle complex data is the MultiIndex, or hierarchical indexing. This tutorial will walk you through what a MultiIndex is and how to effectively create and utilize one.

Understanding MultiIndex

A MultiIndex (also known as hierarchical indexes) allows you to have multiple levels of indexes on a single axis. It is a powerful tool that provides a way to work with higher dimensional data more compactly and efficiently using a two-dimensional table by grouping and accessing data in a more sophisticated manner.

Let’s take a simple example to understand the basic concept:

import pandas as pd
import numpy as np

data = np.random.randn(4, 2)
columns = pd.MultiIndex.from_tuples([('A', 'a'), ('A', 'b')])
df = pd.DataFrame(data, columns=columns)

The output might look like this:

         a         b
0 -0.234566  1.345454
1 -1.003245  0.856423
2  0.234245 -0.956456
3  1.345678 -2.345345

In this example, we have a DataFrame with columns indexed at two levels, ‘A’ at the first level and ‘a’, ‘b’ at the second level. This arrangement allows for organizing and retrieving data in a more structured way.

Creating a MultiIndex

From Arrays or Lists

The easiest way to create a MultiIndex is from arrays or lists. This method involves specifying the levels of indexing directly:

arrays = [[
    'bar', 'bar', 'baz', 'baz',
    'foo', 'foo', 'qux', 'qux'
], [
    'one', 'two', 'one', 'two',
    'one', 'two', 'one', 'two'
index = pd.MultiIndex.from_arrays(arrays, names=('first', 'second'))
df = pd.DataFrame(np.random.randn(8, 4), index=index)

This creates a DataFrame with a hierarchical index named ‘first’ and ‘second’, demonstrating a more complex data structure that allows for segmented operations like slicing and aggregating data at different levels.

From Tuples

Another way to create a MultiIndex is from a list of tuples. Each tuple represents a single level in the hierarchy:

tuples = list(zip(*arrays))
index = pd.MultiIndex.from_tuples(tuples, names=['first', 'second'])
df = pd.DataFrame(np.random.randn(8, 4), index=index)

Using a Product

For indices that represent a product of two sets of labels, you can use pd.MultiIndex.from_product. This method is efficient for creating a grid of indices:

levels = [['foo', 'bar', 'baz', 'qux'], ['one', 'two']]
index = pd.MultiIndex.from_product(levels, names=['first', 'second'])
df = pd.DataFrame(np.random.randn(8, 4), index=index)

Advanced MultiIndex Operations

With a basic understanding of how to create a MultiIndex, let’s explore some advanced operations you can perform using MultiIndex.


DataFrames with MultiIndexes enable advanced slicing operations. You can slice on any level of the hierarchy:

df.loc[('bar', 'one')]


Before performing operations like slicing, it is important to sort your indices. You can easily sort the levels of a MultiIndex DataFrame using sort_index():


Stacking and Unstacking

Stacking and unstacking are powerful techniques to reshape your DataFrame. Stacking converts your columns to rows (make your DataFrame taller), while unstacking converts your rows to columns (make your DataFrame wider). These operations make it easier to analyze complex data:



Throughout this tutorial, we’ve seen how to create and manipulate MultiIndex in Pandas DataFrames, further enhancing your data analysis capabilities. Mastering MultiIndex will allow you to manage complex datasets more effectively and perform sophisticated data analysis and manipulation tasks with ease. Happy coding!