Python – Drop multiple levels from a multi-level column index in Pandas dataframe

When working with Pandas DataFrames that have multi-level column indexes, you can drop multiple levels using the columns.droplevel() method repeatedly. This is useful when you need to simplify complex hierarchical column structures.

Creating a Multi-Level Column Index

First, let's create a DataFrame with a three-level column index using MultiIndex.from_tuples() ?

import numpy as np
import pandas as pd

# Create multi-level column index
items = pd.MultiIndex.from_tuples([("Col 1", "Col 1", "Col 1"),
                                   ("Col 2", "Col 2", "Col 2"),
                                   ("Col 3", "Col 3", "Col 3")])

# Create multi-level row index  
arr = [np.array(['car', 'car', 'car', 'bike', 'bike', 'bike', 'truck', 'truck', 'truck']),
       np.array(['valueA', 'valueB', 'valueC', 'valueA', 'valueB', 'valueC', 'valueA', 'valueB', 'valueC'])]

# Create DataFrame with multi-level indexes
dataFrame = pd.DataFrame(np.random.randn(9, 3), index=arr, columns=items)
dataFrame.index.names = ['vehicle', 'type']

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

Original DataFrame:
                Col 1     Col 2     Col 3
                Col 1     Col 2     Col 3
                Col 1     Col 2     Col 3
vehicle type                              
car     valueA  0.425077  0.020606  1.148156
        valueB -1.720355  0.502863  1.184753
        valueC  0.373106  1.300935 -0.128404
bike    valueA -0.648708  0.944725  0.593327
        valueB -0.613921 -0.238730 -0.218448
        valueC  0.313042 -0.628065  0.910935
truck   valueA  0.286377  0.478067 -1.000645
        valueB  1.151793 -0.171433 -0.612346
        valueC -1.358061  0.735075  0.092700

Dropping the First Level

Use droplevel(0) to remove the topmost level from the column index ?

import numpy as np
import pandas as pd

# Create the same DataFrame as before
items = pd.MultiIndex.from_tuples([("Col 1", "Col 1", "Col 1"),
                                   ("Col 2", "Col 2", "Col 2"),
                                   ("Col 3", "Col 3", "Col 3")])

arr = [np.array(['car', 'car', 'car', 'bike', 'bike', 'bike', 'truck', 'truck', 'truck']),
       np.array(['valueA', 'valueB', 'valueC', 'valueA', 'valueB', 'valueC', 'valueA', 'valueB', 'valueC'])]

dataFrame = pd.DataFrame(np.random.randn(9, 3), index=arr, columns=items)
dataFrame.index.names = ['vehicle', 'type']

# Drop the first level
dataFrame.columns = dataFrame.columns.droplevel(0)
print("After dropping first level:")
print(dataFrame)

After dropping first level:
                Col 1     Col 2     Col 3
                Col 1     Col 2     Col 3
vehicle type                              
car     valueA  0.425077  0.020606  1.148156
        valueB -1.720355  0.502863  1.184753
        valueC  0.373106  1.300935 -0.128404
bike    valueA -0.648708  0.944725  0.593327
        valueB -0.613921 -0.238730 -0.218448
        valueC  0.313042 -0.628065  0.910935
truck   valueA  0.286377  0.478067 -1.000645
        valueB  1.151793 -0.171433 -0.612346
        valueC -1.358061  0.735075  0.092700

Dropping Multiple Levels

After dropping one level, the remaining levels shift down. Drop another level to get a single-level column index ?

import numpy as np
import pandas as pd

# Create the same DataFrame
items = pd.MultiIndex.from_tuples([("Col 1", "Col 1", "Col 1"),
                                   ("Col 2", "Col 2", "Col 2"),
                                   ("Col 3", "Col 3", "Col 3")])

arr = [np.array(['car', 'car', 'car', 'bike', 'bike', 'bike', 'truck', 'truck', 'truck']),
       np.array(['valueA', 'valueB', 'valueC', 'valueA', 'valueB', 'valueC', 'valueA', 'valueB', 'valueC'])]

dataFrame = pd.DataFrame(np.random.randn(9, 3), index=arr, columns=items)
dataFrame.index.names = ['vehicle', 'type']

# Drop first level
dataFrame.columns = dataFrame.columns.droplevel(0)

# Drop second level (now at index 0)
dataFrame.columns = dataFrame.columns.droplevel(0)

print("After dropping two levels:")
print(dataFrame)

After dropping two levels:
                Col 1     Col 2     Col 3
vehicle type                              
car     valueA  0.425077  0.020606  1.148156
        valueB -1.720355  0.502863  1.184753
        valueC  0.373106  1.300935 -0.128404
bike    valueA -0.648708  0.944725  0.593327
        valueB -0.613921 -0.238730 -0.218448
        valueC  0.313042 -0.628065  0.910935
truck   valueA  0.286377  0.478067 -1.000645
        valueB  1.151793 -0.171433 -0.612346
        valueC -1.358061  0.735075  0.092700

Alternative: Drop Specific Level by Position

You can also drop a specific level by its position without dropping sequentially ?

import numpy as np
import pandas as pd

# Create DataFrame with 3-level columns
items = pd.MultiIndex.from_tuples([("Level0", "Level1", "Col1"),
                                   ("Level0", "Level1", "Col2"), 
                                   ("Level0", "Level1", "Col3")])

data = [[1, 2, 3], [4, 5, 6]]
df = pd.DataFrame(data, columns=items)

print("Original:")
print(df)

# Drop level 1 (middle level)
df.columns = df.columns.droplevel(1)
print("\nAfter dropping level 1:")
print(df)

Original:
  Level0 Level0 Level0
  Level1 Level1 Level1
    Col1   Col2   Col3
0      1      2      3
1      4      5      6

After dropping level 1:
  Level0 Level0 Level0
    Col1   Col2   Col3
0      1      2      3
1      4      5      6

Conclusion

Use columns.droplevel() repeatedly to remove multiple levels from multi-level column indexes. Remember that after each drop, the remaining levels shift positions, so always drop from index 0 for sequential removal.