Pandas DataFrame | nunique method
Start your free 7-days trial now!
Pandas DataFrame.nunique(~) method counts the number of unique values for each row or column in the DataFrame.
Parameters
1. axislink | int or string
The axis along which to count the number of unique values:
Value | Meaning |
|---|---|
| Count each column. |
| Count each row. |
By default, axis=0.
2. dropnalink | boolean | optional
Whether or not to ignore NaN. By default, dropna=True.
Return Value
A Series that holds the count of unique numbers in each row or column of the source DataFrame.
Examples
Consider the following DataFrame:
df
A B C0 5 2 11 5 2 22 5 3 4
Counting column-wise
To count the number of unique numbers in each column:
df.nunique() # axis=0
A 1B 2C 3dtype: int64
This tells us that:
column
Ahas 1 unique value.column
Bhas 2 unique values.column
Chas 3 unique values.
Counting row-wise
To count the number of unique numbers in each row, set axis=1:
df.nunique(axis=1)
0 21 22 3dtype: int64
This tells us that:
row
0has 2 unique values.row
1has 2 unique values.row
2has 3 unique values.
Dealing with missing values
Consider the following DataFrame with two missing values:
df
A0 NaN1 1.02 NaN
By default, dropna=True, which means that NaNs will be ignored:
df.nunique()
A 1dtype: int64
We can consider them as values by setting dropna=False:
df.nunique(dropna=False)
A 2dtype: int64
This tell us that column A has 2 unique values: NaN and 1.0.
