Working with MultiIndex DataFrames

Guide to creating, inspecting, and working with MultiIndex DataFrames in pandas, including integration with NumPy.

Working with MultiIndex DataFrames

This notebook covers MultiIndex DataFrames in pandas: creation, inspection, and combining with NumPy arrays. MultiIndex allows hierarchical indexing for complex data structures.

Introduction to MultiIndex

MultiIndex in pandas allows you to have multiple levels of indexing on rows or columns. It’s useful for hierarchical data like time series with multiple categories.

import pandas as pd

Creating a MultiIndex DataFrame

Use pd.MultiIndex.from_arrays() to create a MultiIndex from arrays. Here, we create a DataFrame with a two-level row index.

arrays = [['A','A','B','B',], [1,2,1,2]]
index = pd.MultiIndex.from_arrays(arrays, names=('First', 'Second'))
df = pd.DataFrame({'Data': [10,20,30,40]}, index=index)

df

		Data
First	Second
A	1	10
	2	20
B	1	30
	2	40

Inspecting the MultiIndex

Access the index with df.index. It shows the hierarchical structure.

df.index

MultiIndex([('A', 1),
            ('A', 2),
            ('B', 1),
            ('B', 2)],
           names=['First', 'Second'])

Combining pandas with NumPy

Pandas integrates seamlessly with NumPy. You can create DataFrames from NumPy arrays and use NumPy functions on DataFrame data.

Creating DataFrames from NumPy Arrays

Use pd.DataFrame() with a NumPy array to create a DataFrame. Specify column names for clarity.

import numpy as np

array_data = np.array([[1,2,3], [4,5,6], [7,8,9]])
df_np = pd.DataFrame(array_data, columns=['Sales A', 'Sales B', 'Sales C'])

df_np

	Sales A	Sales B	Sales C
0	1	2	3
1	4	5	6
2	7	8	9

Best Practices

• Use meaningful names for MultiIndex levels (e.g., ‘Category’, ‘Subcategory’).
• When selecting data, use .loc[] with tuples for MultiIndex access.
• Reset index with df.reset_index() if you need to flatten the hierarchy.

Summary

This notebook demonstrated creating and inspecting MultiIndex DataFrames and integrating pandas with NumPy arrays. MultiIndex is powerful for complex data but can be tricky—practice with real datasets!