Stack and unstack in Data Analysis Python - Step-by-Step Execution

Concept Flow - Stack and unstack

Start with DataFrame

↓

Apply stack()

↓

DataFrame becomes Series with MultiIndex

↓

Apply unstack()

↓

Series with MultiIndex becomes DataFrame

↓

End

Stack turns columns into rows creating a Series with MultiIndex. Unstack reverses this, turning rows back into columns.

Execution Sample

Data Analysis Python

import pandas as pd

df = pd.DataFrame({
    'A': [1, 2],
    'B': [3, 4]
})
stacked = df.stack()
unstacked = stacked.unstack()

This code creates a simple DataFrame, stacks it to convert columns into rows, then unstacks it back to original form.

Execution Table

Step	Action	Data Structure Type	Shape/Index	Sample Data
1	Create DataFrame	DataFrame	(2 rows, 2 columns)	{(0, 'A'):1, (0, 'B'):3, (1, 'A'):2, (1, 'B'):4}
2	Apply stack()	Series with MultiIndex	(4 elements)	{(0, 'A'):1, (0, 'B'):3, (1, 'A'):2, (1, 'B'):4}
3	Apply unstack()	DataFrame	(2 rows, 2 columns)	{(0, 'A'):1, (0, 'B'):3, (1, 'A'):2, (1, 'B'):4}
4	End	DataFrame	(2 rows, 2 columns)	Same as Step 1

💡 Unstack reverses stack, restoring original DataFrame shape and data.

Variable Tracker

Variable	Start	After stack()	After unstack()
df	DataFrame with 2 rows, 2 columns	Unchanged	Unchanged
stacked	Not defined	Series with MultiIndex, 4 elements	Unchanged
unstacked	Not defined	Not defined	DataFrame same as df

Key Moments - 2 Insights

Why does stack() result in a Series with MultiIndex instead of a DataFrame?

Stack compresses columns into a single index level, so the result is a Series indexed by the original row index plus the stacked column labels, forming a MultiIndex (see execution_table Step 2).

Does unstack() always restore the original DataFrame exactly?

Unstack reverses stack only if the MultiIndex matches the original structure. If the Series has missing or extra index levels, unstack may produce a different shape (see execution_table Step 3).

Visual Quiz - 3 Questions

Look at the execution_table, what is the data structure type after applying stack()?

A DataFrame

B Series with MultiIndex

C List

D Dictionary

💡 Hint

Check Step 2 in execution_table under 'Data Structure Type'

At which step does the data shape change from (2 rows, 2 columns) to a Series with 4 elements?

A Step 2

B Step 1

C Step 3

D Step 4

💡 Hint

Look at the 'Shape/Index' column in execution_table for Step 2

If the stacked Series had missing values, how would unstack() affect the final DataFrame?

A It would produce an error

B It would remove rows with missing values

C It would fill missing values with NaN in the DataFrame

D It would ignore missing values and produce the same DataFrame

💡 Hint

Unstack creates a DataFrame with NaN where data is missing, see pandas behavior for unstack

Concept Snapshot

Stack and unstack:
- stack() converts columns into rows, creating a Series with MultiIndex.
- unstack() reverses stack, turning MultiIndex Series back into a DataFrame.
- Useful for reshaping data between wide and long formats.
- stack() reduces columns, unstack() expands them.
- Missing data after stack leads to NaN after unstack.

Full Transcript

Stack and unstack are pandas methods to reshape data. Starting with a DataFrame, stack() compresses columns into rows, resulting in a Series with a MultiIndex combining original row and column labels. This changes the shape from a 2D table to a 1D Series with hierarchical indexing. Unstack() reverses this process, taking a MultiIndex Series and expanding it back into a DataFrame with columns restored. The execution trace shows creating a DataFrame with two columns and two rows, stacking it to get a Series with four elements indexed by row and column labels, then unstacking back to the original DataFrame shape. Key points include understanding that stack changes the data structure type and shape, and unstack can restore the original DataFrame if the index matches. Missing data in the stacked Series will appear as NaN in the unstacked DataFrame. This process is useful for converting data between wide and long formats in data analysis.