Drop Columns with NaN Values in Pandas DataFrame

Here are 2 ways to drop columns with NaN values in Pandas DataFrame:

(1) Drop any column that contains at least one NaN:

df.dropna(axis='columns', inplace=True)

(2) Drop column/s where ALL the values are NaN:

df.dropna(axis='columns', how='all', inplace=True)

The Example

For demonstration purposes, let’s create a DataFrame with 5 columns, where:

3 columns will contain NaN values
2 columns will not contain any NaN values

Here is the syntax to create the DataFrame:

import pandas as pd
import numpy as np

data = {'Column_A': [1, 2, 3, 4, 5, np.nan, 6, 7, np.nan],
        'Column_B': [11, 22, 33, 44, 55, 66, 77, 88, 99],
        'Column_C': ['a', 'b', np.nan, np.nan, 'c', 'd', 'e', np.nan, 'f'],
        'Column_D': ['aa', 'bb', 'cc', 'dd', 'ee', 'ff', 'gg', 'hh', 'ii'],
        'Column_E': [np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan]
        }

df = pd.DataFrame(data)

print(df)

As you can see, 3 columns (‘Column_A’, ‘Column_C’ and ‘Column_E’) contain NaN values:

   Column_A  Column_B Column_C Column_D  Column_E
0       1.0        11        a       aa       NaN
1       2.0        22        b       bb       NaN
2       3.0        33      NaN       cc       NaN
3       4.0        44      NaN       dd       NaN
4       5.0        55        c       ee       NaN
5       NaN        66        d       ff       NaN
6       6.0        77        e       gg       NaN
7       7.0        88      NaN       hh       NaN
8       NaN        99        f       ii       NaN

The ultimate goal is to drop the columns with the NaN values in the above DataFrame.

(1) Drop any column that contains at least one NaN

You can use the following template to drop any column that contains at least one NaN:

df.dropna(axis='columns', inplace=True)

For our example:

import pandas as pd
import numpy as np

data = {'Column_A': [1, 2, 3, 4, 5, np.nan, 6, 7, np.nan],
        'Column_B': [11, 22, 33, 44, 55, 66, 77, 88, 99],
        'Column_C': ['a', 'b', np.nan, np.nan, 'c', 'd', 'e', np.nan, 'f'],
        'Column_D': ['aa', 'bb', 'cc', 'dd', 'ee', 'ff', 'gg', 'hh', 'ii'],
        'Column_E': [np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan]
        }

df = pd.DataFrame(data)

df.dropna(axis='columns', inplace=True)

print(df)

Once you run the code, you’ll notice that the 3 columns, which originally contained the NaN values, were dropped. Only the other 2 columns (without the NaN values) were maintained:

   Column_B  Column_D
0        11        aa
1        22        bb
2        33        cc
3        44        dd
4        55        ee
5        66        ff
6        77        gg
7        88        hh
8        99        ii

(2) Drop column/s where ALL the values are NaN

What if you’d like to drop only the column/s where ALL the values are NaN?

In that case, you can use the template below to accomplish this goal:

df.dropna(axis='columns', how='all', inplace=True)

Here is the code that you can use:

import pandas as pd
import numpy as np

data = {'Column_A': [1, 2, 3, 4, 5, np.nan, 6, 7, np.nan],
        'Column_B': [11, 22, 33, 44, 55, 66, 77, 88, 99],
        'Column_C': ['a', 'b', np.nan, np.nan, 'c', 'd', 'e', np.nan, 'f'],
        'Column_D': ['aa', 'bb', 'cc', 'dd', 'ee', 'ff', 'gg', 'hh', 'ii'],
        'Column_E': [np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, np.nan]
        }

df = pd.DataFrame(data)

df.dropna(axis='columns', how='all', inplace=True)

print(df)

As you may notice, ‘Column_E‘ (that contained only NaN values) was dropped:

   Column_A  Column_B  Column_C  Column_D
0       1.0        11         a       aa
1       2.0        22         b       bb
2       3.0        33       NaN       cc
3       4.0        44       NaN       dd
4       5.0        55         c       ee
5       NaN        66         d       ff
6       6.0        77         e       gg
7       7.0        88       NaN       hh
8       NaN        99         f       ii

Additional Resources

You can check the Pandas Documentation to learn more about dropna.

Optionally, you can check the following guide to learn how to drop rows with NaN values in Pandas DataFrame.