How to use equivalent of np.where() in Pandas

---

You can use the NumPy function to quickly update the values in a NumPy array using if-else logic.

For example, the following code shows how to update the values in a NumPy array that meet a certain condition:

import numpy as np

#create NumPy array of values
x = np.array([1, 3, 3, 6, 7, 9])

#update valuesin array based on condition
x = np.where((x < 5) | (x > 8), x/2, x)

#view updated array
x

array([0.5, 1.5, 1.5, 6. , 7. , 4.5])

If a given value in the array was less than 5 or greater than 8, we divided the value by 2.

Else, we left the value unchanged.

We can perform a similar operation in a pandas DataFrame by using the pandas function, but the syntax is slightly different.

Here’s the basic syntax using the NumPy where() function:

x = np.where(condition, value_if_true, value_if_false)

And here’s the basic syntax using the pandas where() function:

df['col'] = (value_if_false).where(condition, value_if_true)

The following example shows how to use the pandas where() function in practice.

Example: The Equivalent of np.where() in Pandas

Suppose we have the following pandas DataFrame:

import pandas as pd

#create DataFrame
df = pd.DataFrame({'A': [18, 22, 19, 14, 14, 11, 20, 28],
                   'B': [5, 7, 7, 9, 12, 9, 9, 4]})

#view DataFrame
print(df)

    A   B
0  18   5
1  22   7
2  19   7
3  14   9
4  14  12
5  11   9
6  20   9
7  28   4

We can use the following pandas where() function to update the values in column A based on a specific condition:

#update values in column A based on condition
df['A'] = (df['A'] / 2).where(df['A'] < 20, df['A'] * 2)

#view updated DataFrame
print(df)

      A   B
0   9.0   5
1  44.0   7
2   9.5   7
3   7.0   9
4   7.0  12
5   5.5   9
6  40.0   9
7  56.0   4

If a given value in column A was less than 20, we multiplied the value by 2.