Rename Column Name(s) in Pandas DataFrame

When working with datasets from external sources - column names can get wild. Different naming conventions, cases (snake_case, CamelCase, etc.), as well as names are common. A common headache is caused by really long column names, that you might have to call on many times in the lifecycle of data manipulation.

One of the first steps in many preprocessing pipelines is renaming the columns.

Let's create a simple mock DataFrame with a couple of columns:

import pandas as pd

df = pd.DataFrame({'Short col name': [1], 
                  'Really long column name': [2],
                   'col3': [3]})
                   
print(df)

It has three columns with one row each:

Short col name  Really long column name  col3
0               1                        2     3

Rename Specific Columns

To rename the columns, you call the DataFrame.rename() method, providing the columns argument with a dictionary of {'old name' : 'new name'}:

df = df.rename(columns={'Short col name': 'col1', 'Really long column name': 'col2'})

print(df)

This results in:

   col1  col2  col3
0     1     2     3

Note: You don't need to provide every column here - we've totally skipped col3 because it already follows our mock convention. Only the columns you reference will be found and renamed.

This operation is not done in-place, so you'll want to assign the result of the method to a new DataFrame instance or the object already in memory as we have. To perform the operation in-place, add the inplace flag:

df.rename(inplace=True, columns={'Short col name': 'col1', 'Really long column name': 'col2'})
print(df)

This results in:

  col1  col2  col3
0     1     2     3

Rename All Columns

Another way to rename the columns is to, well, replace the columns (list of column names) by pointing it to another array:

df.columns = ['col_1', 'col_2', 'col_3']
print(df)

This results in:

   col_1  col_2  col_3
0      1      2      3