How to Use Pandas apply() inplace?

The apply() function in pandas is used to apply a custom function to a DataFrame or Series. By default, apply() returns a new DataFrame or Series, but you can modify the original data in-place using specific techniques. In this article, we'll explore how to achieve in-place modifications with apply().

Syntax

DataFrame.apply(func, axis=0)
Series.apply(func)

The axis parameter determines whether to apply the function row-wise (axis=1) or column-wise (axis=0). The func can be a built-in function, lambda function, or custom function.

Default Behavior ? Creating New Objects

When using apply() without in-place assignment, it returns a new DataFrame or Series, leaving the original unchanged −

import pandas as pd

# Create a DataFrame
data = {'Name': ['John', 'Emily', 'James', 'Emma'],
        'Age': [28, 32, 25, 29]}
df = pd.DataFrame(data)

# Function to add a prefix to names
def add_prefix(name):
    return 'Mr. ' + name

# Apply function without modifying original
new_series = df['Name'].apply(add_prefix)
print("Original DataFrame:")
print(df)
print("\nNew Series:")
print(new_series)

Original DataFrame:
    Name  Age
0   John   28
1  Emily   32
2  James   25
3   Emma   29

New Series:
0     Mr. John
1    Mr. Emily
2    Mr. James
3     Mr. Emma
Name: Name, dtype: object

Achieving In-Place Modification

To modify the original DataFrame in-place, assign the result of apply() back to the column or DataFrame −

import pandas as pd

# Create a DataFrame
data = {'Name': ['John', 'Emily', 'James', 'Emma'],
        'Age': [28, 32, 25, 29]}
df = pd.DataFrame(data)

# Function to add a prefix to names
def add_prefix(name):
    return 'Mr. ' + name

# Modify original DataFrame in-place
df['Name'] = df['Name'].apply(add_prefix)

print("Modified DataFrame:")
print(df)

Modified DataFrame:
       Name  Age
0   Mr. John   28
1  Mr. Emily   32
2  Mr. James   25
3   Mr. Emma   29

Working with Multiple Columns

You can apply functions to multiple columns simultaneously using axis=1 for row-wise operations −

import pandas as pd

# Create a DataFrame
data = {'Name': ['John', 'Emily', 'James', 'Emma'],
        'Age': [28, 32, 25, 29],
        'Salary': [50000, 60000, 45000, 55000]}
df = pd.DataFrame(data)

# Function to process multiple columns
def process_data(row):
    row['Name'] = 'Mr. ' + row['Name']
    row['Salary'] *= 2
    return row

# Apply to specific columns in-place
df[['Name', 'Salary']] = df[['Name', 'Salary']].apply(process_data, axis=1)

print("Modified DataFrame:")
print(df)

Modified DataFrame:
        Name  Age  Salary
0   Mr. John   28  100000
1  Mr. Emily   32  120000
2  Mr. James   25   90000
3   Mr. Emma   29  110000

Memory and Performance Considerations

Important: In-place modification overwrites original data permanently. Always backup important data before applying in-place operations, especially when working with large datasets where memory efficiency is crucial.

Conclusion

While pandas apply() doesn't have a built-in inplace parameter, you can achieve in-place modifications by assigning results back to the original DataFrame. Use df['column'] = df['column'].apply(func) for single columns or select multiple columns for batch operations.