Pandas DataFrame mask() Method

In pandas, the mask() method is used to replace values in a DataFrame or Series where a specified condition is True. It essentially allows you to mask or hide specific data based on a condition and replace it with another value.

Pandas DataFrame mask() Introduction

Let’s know the syntax of the mask() method.


# Syntax of DataFrame mask()
DataFrame.mask(cond, other=nan, inplace=False, axis=None, level=None, errors='raise')

Parameters of the DataFrame mask()

Following are the parameters of the DataFrame mask() method.

cond – A boolean condition or DataFrame of the same shape as the DataFrame being masked. Values in the DataFrame where the condition is True will be replaced.
other – The value to replace entries that match the condition. By default, this is NaN.
inplace – If True, performs the operation in place and returns None. Defaults to False.
axis – The axis to align the condition with (0 for rows, 1 for columns). Defaults to None.
level – For MultiIndex DataFrames, specifies the level at which to perform the operation. Defaults to None.
errors – Controls whether to raise errors if the conditions cannot be applied. Defaults to raise.

Return Value

It returns a DataFrame with values replaced according to the condition specified.

Usage of Pandas DataFrame mask() Method

The mask() method in pandas is used for the conditional replacement of values in a DataFrame.

Now, Let’s create Pandas DataFrame using data from a Python dictionary, where the columns are A, and B.


import pandas as pd

# Create a DataFrame
df = pd.DataFrame({
    'A': [5, 8, 13, 15],
    'B': [2, 12, 4, 7]
})

print("Original DataFrame:\n",df)

Yields below output

Here is an example demonstrating the basic usage of the mask() method in Pandas to replace values in a DataFrame based on a specified condition.


# Mask values greater than 10
df2 = df.mask(df > 10)
print("Masked DataFrame (values > 10 replaced with NaN):\n", df2)

In the above example, any values greater than 10 in the DataFrame are replaced with NaN.

Custom Replacement Value

Alternatively, to use a custom replacement value with the mask() method, you can specify the other parameter.


# Mask values greater than 10 and replace with 0
df2 = df.mask(df > 10, other=0)
print("Masked values > 10 replaced with 0:\n", df2)

# Output:
# Masked values > 10 replaced with 0:
#    A  B
# 0  5  2
# 1  8  0
# 2  0  4
# 3  0  7

In the above example, any values in the DataFrame greater than 10 are replaced with 0 instead of NaN.

Using inplace Parameter

To use the inplace parameter with the mask() method, you can modify the original DataFrame directly without creating a new one. Setting inplace=True applies the mask and updates the DataFrame in place.

Here’s how you can use the inplace parameter to mask values greater than 10 and replace them with -6.


# Mask values greater than 10 and replace with -6, modifying in place
df.mask(df > 10, other=-6, inplace=True)
print("Modified DataFrame (values > 10 replaced with -6):\n", df)

# Output:
# Modified DataFrame (values > 10 replaced with -6):
#     A  B
# 0  5  2
# 1  8 -6
# 2 -6  4
# 3 -6  7

In the above example, the inplace=True parameter modifies the original DataFrame df directly, replacing any values greater than 10 with -6.

Applying Mask on Specific Axis

Similarly, to apply the mask() method on a specific axis of a DataFrame, you can use the axis parameter. This parameter allows you to specify whether the mask should be applied along rows (axis=0) or columns (axis=1).

Masking Along Columns (axis=1)

You can use the mask() method in Pandas to replace values greater than 5 across columns (axis=1) with NaN.


# Mask values greater than 5 along columns axis
df2 = df.mask(df > 5, axis=1)
print("Masked values > 5 replaced with NaN along columns:\n", df2)

# Output:
# Masked values > 5 replaced with NaN along columns:
#      A    B
# 0  5.0  2.0
# 1  NaN  NaN
# 2  NaN  4.0
# 3  NaN  NaN

In the above example, values greater than 5 are replaced with NaN along the columns axis, meaning each value is checked and replaced based on its column.

Masking Along Rows (axis=0)

You can also use the mask() method in Pandas to replace values greater than 5 across rows (axis=0) with NaN.


# Mask values greater than 5 along rows axis
df2 = df.mask(df > 5, axis=0)
print("Masked values > 5 replaced with NaN along rows:\n", df2)

# Output:
# Masked values > 5 replaced with NaN along rows:
#      A    B
# 0  5.0  2.0
# 1  NaN  NaN
# 2  NaN  4.0
# 3  NaN  NaN

In the above example, values greater than 5 are replaced with NaN along the rows axis, meaning each value is checked and replaced based on its row.

Using a Callable Condition

Finally, you can also use the mask() method with a callable condition in Pandas. For instance, a condition function can be used to identify even values in the DataFrame. The mask() method will then replace those even values with NaN.


# Define a callable condition
def condition(x):
    return x % 2 == 0

# Mask values based on the callable condition
df2 = df.mask(condition)
print("Masked DataFrame (even values replaced with NaN):\n", df2)

# Output:
# Masked DataFrame (even values replaced with NaN):
#        A    B
# 0   5.0  NaN
# 1   NaN  NaN
# 2  13.0  NaN
# 3  15.0  7.0

Frequently Asked Questions on Pandas DataFrame mask() Method

What is the purpose of the mask() method in Pandas?

The mask() method in Pandas is used to replace values in a DataFrame or Series where a specified condition is true. It allows for conditional data replacement, enabling you to mask or hide certain values based on the condition.

Can you replace masked values with a custom value instead of NaN?

You can replace masked values with a custom value instead of NaN by using the other parameter in the mask() method.

Is it possible to apply the mask() method on a specific axis?

You can specify the axis along which to apply the mask() method using the axis parameter.

Can the mask() method be used with MultiIndex DataFrames?

The mask() method supports MultiIndex DataFrames and allows specifying the level parameter to apply the condition on a particular level of the index.

How does the mask() method differ from the where() method in Pandas?

The mask() method replaces values where the condition is True, whereas the where() method replaces values where the condition is False. They are essentially complementary methods used for conditional replacement in Pandas.

Conclusion

In conclusion, the mask() method in Pandas is a versatile and powerful tool for conditional data manipulation, allowing you to replace values in a DataFrame or Series based on specific conditions. You can replace values with NaN by default or specify a custom replacement value using the other parameter. Additionally, you can modify the DataFrame in place with the inplace parameter, and you can apply conditions to specific axes or use callable conditions for more complex scenarios.

Happy Learning!!

Reference

https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.where.html