Python – Stacking a single-level column with Pandas stack()?

The Pandas stack() method transforms a DataFrame by stacking column levels into row levels, creating a hierarchical index. This operation pivots columns into a multi-level index, converting wide data to long format.

Syntax

DataFrame.stack(level=-1, dropna=True)

Creating a DataFrame with Single-Level Columns

First, let's create a simple DataFrame with single-level columns ?

import pandas as pd

# Create DataFrame with single-level columns
dataFrame = pd.DataFrame([[10, 15], [20, 25], [30, 35], [40, 45]],
                        index=['w', 'x', 'y', 'z'],
                        columns=['a', 'b'])

print("Original DataFrame:")
print(dataFrame)

Original DataFrame:
    a   b
w  10  15
x  20  25
y  30  35
z  40  45

Stacking the DataFrame

The stack() method converts columns to rows, creating a Series with a multi-level index ?

import pandas as pd

# Create DataFrame
dataFrame = pd.DataFrame([[10, 15], [20, 25], [30, 35], [40, 45]],
                        index=['w', 'x', 'y', 'z'],
                        columns=['a', 'b'])

# Stack the DataFrame
stacked = dataFrame.stack()
print("Stacked DataFrame:")
print(stacked)
print(f"\nType: {type(stacked)}")

Stacked DataFrame:
w  a    10
   b    15
x  a    20
   b    25
y  a    30
   b    35
z  a    40
   b    45
dtype: int64

Type: <class 'pandas.core.series.Series'>

Understanding the Result

The stacked result has a two-level index where the first level is the original row index and the second level is the original column names ?

import pandas as pd

dataFrame = pd.DataFrame([[10, 15], [20, 25], [30, 35], [40, 45]],
                        index=['w', 'x', 'y', 'z'],
                        columns=['a', 'b'])

stacked = dataFrame.stack()

# Access the multi-level index
print("Index levels:")
print(f"Level 0 (rows): {stacked.index.get_level_values(0).tolist()}")
print(f"Level 1 (cols): {stacked.index.get_level_values(1).tolist()}")

# Access specific values
print(f"\nValue at (w, a): {stacked['w', 'a']}")
print(f"Value at (x, b): {stacked['x', 'b']}")

Index levels:
Level 0 (rows): ['w', 'w', 'x', 'x', 'y', 'y', 'z', 'z']
Level 1 (cols): ['a', 'b', 'a', 'b', 'a', 'b', 'a', 'b']

Value at (w, a): 10
Value at (x, b): 25

Key Points

• The stack() method returns a Series with a hierarchical index
• Original column names become the second level of the new index
• The operation is the inverse of unstack()
• By default, dropna=True removes missing values

Conclusion

The stack() method is essential for reshaping data from wide to long format. It creates a hierarchical index by moving column labels to row levels, making it useful for data analysis and visualization tasks that require long-format data.