Count Unique Values in all Columns of Pandas Dataframe

This article will discuss different ways to count unique values in all columns of a dataframe in Pandas.

First of all, we will create a sample Dataframe from a list of tuples i.e.

import pandas as pd
import numpy as np

# List of Tuples
list_of_tuples = [  (11,  34, 67,       5,  np.NaN, 34),
                    (12,  34, np.NaN,   11, 12,     np.NaN),
                    (13,  34, 78,       11, 12,     18) ,
                    (12,  34, 80,       41, 11,     18)]


# Create a DataFrame object
df = pd.DataFrame(  list_of_tuples, 
                    columns=['A', 'B', 'C', 'D', 'E', 'F'])

print(df)

Contents of the Dataframe are,

    A   B     C   D     E     F
0  11  34  67.0   5   NaN  34.0
1  12  34   NaN  11  12.0   NaN
2  13  34  78.0  11  12.0  18.0
3  12  34  80.0  41  11.0  18.0

Now let’s see how we can get the count of unique values in each of the columns.

Count Unique Values in All Columns using Dataframe.nunique()

In Pandas, the Dataframe provides a member function nunique(). It gives a Series containing unique elements along the requested axis. We can use this to get a count of unique values in each of the columns. For example,

# Get a Series of count of unique values in each column
unique_values = df.nunique()

print(unique_values)

Output:

A    3
B    1
C    3
D    3
E    2
F    2
dtype: int64

Here we fetched the count of unique values in each of the columns of Dataframe.

By default Dataframe.nunique() doesn’t includes the NaN values. Therefore, if you want to include the NaN values while counting unique values, you need to pass the dropna argument as False to the nunique() function. For example,

# Get a Series of count of unique values in each column
# including NaN
unique_values = df.nunique(dropna=False)

print(unique_values)

Output:

A    3
B    1
C    4
D    3
E    3
F    3
dtype: int64

This time nunique() included the NaN values to while counting the unique elements.

Count Unique Values in All Columns using For Loop

Another simple solution is that we can iterate over all the columns of a Datframe one by one. During iteration, we can count the unique values of each column. For example,

# Iterate over all column names of Dataframe
for col in df.columns:
    # Select the column by name and get count of unique values in it
    count = df[col].nunique()
    print('Count of Unique values in Column ', col, ' is : ', count)

Output:

Count of Unique values in Column  A  is :  3
Count of Unique values in Column  B  is :  1
Count of Unique values in Column  C  is :  3
Count of Unique values in Column  D  is :  3
Count of Unique values in Column  E  is :  2
Count of Unique values in Column  F  is :  2

Using Loop, we iterated through all the column names of Dataframe. Then for each column name, we fetched the column as a Series object and then counted the unique values in that column using Series.nunique() function.

By default Series.nunique() doesn’t includes the NaN values. Therefore, if you want to include the NaN values while counting unique values, you need to pass the dropna argument as False to the Series.nunique() function. For example,

# Iterate over all column names of Dataframe and Include NaN
for col in df.columns:
    # Select the column by name and get count of unique values in it
    # including NaN
    count = df[col].nunique(dropna=False)
    print('Count of Unique values in Column ', col, ' is : ', count)

Output:

Count of Unique values in Column  A  is :  3
Count of Unique values in Column  B  is :  1
Count of Unique values in Column  C  is :  4
Count of Unique values in Column  D  is :  3
Count of Unique values in Column  E  is :  3
Count of Unique values in Column  F  is :  3

In this way, if you want, you can also skip specific columns based on conditions.

Read More

• Count unique value in a single Dataframe Column
• Pandas – Series.unique() method
• Pandas – Series.nunique() method
• Pandas – Series.is_unique method

The complete example is as follows,

import pandas as pd
import numpy as np

# List of Tuples
list_of_tuples = [  (11,  34, 67,       5,  np.NaN, 34),
                    (12,  34, np.NaN,   11, 12,     np.NaN),
                    (13,  34, 78,       11, 12,     18) ,
                    (12,  34, 80,       41, 11,     18)]


# Create a DataFrame object
df = pd.DataFrame(  list_of_tuples, 
                    columns=['A', 'B', 'C', 'D', 'E', 'F'])

print(df)

# Get a Series of count of unique values in each column
unique_values = df.nunique()

print(unique_values)

print('***********')

# Get a Series of count of unique values in each column
# including NaN
unique_values = df.nunique(dropna=False)

print(unique_values)

print('***********')

# Iterate over all column names of Dataframe
for col in df.columns:
    # Select the column by name and get count of unique values in it
    count = df[col].nunique()
    print('Count of Unique values in Column ', col, ' is : ', count)

print('***********')

# Iterate over all column names of Dataframe and Include NaN
for col in df.columns:
    # Select the column by name and get count of unique values in it
    # including NaN
    count = df[col].nunique(dropna=False)
    print('Count of Unique values in Column ', col, ' is : ', count)

Output:

    A   B     C   D     E     F
0  11  34  67.0   5   NaN  34.0
1  12  34   NaN  11  12.0   NaN
2  13  34  78.0  11  12.0  18.0
3  12  34  80.0  41  11.0  18.0
A    3
B    1
C    3
D    3
E    2
F    2
dtype: int64
***********
A    3
B    1
C    4
D    3
E    3
F    3
dtype: int64
***********
Count of Unique values in Column  A  is :  3
Count of Unique values in Column  B  is :  1
Count of Unique values in Column  C  is :  3
Count of Unique values in Column  D  is :  3
Count of Unique values in Column  E  is :  2
Count of Unique values in Column  F  is :  2
***********
Count of Unique values in Column  A  is :  3
Count of Unique values in Column  B  is :  1
Count of Unique values in Column  C  is :  4
Count of Unique values in Column  D  is :  3
Count of Unique values in Column  E  is :  3
Count of Unique values in Column  F  is :  3

Summary:

We learned two different ways to count unique values in all columns of the Dataframe in Pandas.