Pandas Where will replace values where your condition is False. It is useful when you have values that do not meet a criteria, and they need replacing.
We don’t often use this function, but it can be a handy one liner instead of iterating through a DataFrame or Series with .apply().
1. pd.DataFrame.where(cond=condition_to_check, other="Value To Fill")
Pseudo code: Where a condition is false, replace a value.
Pandas Where
Where
.where() has two main parameters, cond and other. Both of these are flexible to take Series, DataFrame or callable.
- cond: Which stands for condition. This can either be a Series, DataFrame, or callable (function). WHERE this condition is
false, pandas will replace values. I know, it’s a bit counter intuitive. - Other What to replace your values with. This again can be a scalar (single value), Series, DataFrame, or callable.
- Other Parameters: To check out the other parameters that are not used that often, see the official documentation.
Let’s run through each iteration of the above parameters
import pandas as pd
import numpy as np
Pandas Where¶
We will run through 3 examples:
- Pandas Where with DataFrame condition
- Pandas Where with Series condition
- Pandas Where with Callable condition
- Pandas Where with Callable other
- Pandas Where with DataFrame other
But first, let's create our DataFrame of students and test scores
np.random.seed(seed=42)
df = pd.DataFrame(data=np.random.randint(0,100,(4,3)),
columns=('Test1', 'Test2', 'Test3'),
index=['Bob', 'Sally', 'Frank', 'Patty']
)
df
1. Pandas Where with DataFrame condition¶
To start off, let's use a DataFrame in our condition parameter. I want to see all test scores that are above 90. This means to evaluate the scores as FALSE I need to look for all scores BELOW 90.
I'll replace the >90 with "A+." Notice how I first evaluate the DataFrame, check out where the FALSES are. These will get replaced.
df<90
df.where(df<90, "A+")
2. Pandas Where with Series condition¶
Say you wanted to fill an entire row based off of the condition of a single Pandas Series. You can also pass a series as the conditional.
Here, I'll replace a all values in the row where Test3<90
df['Test3']<90
df.where(df['Test3']<90, "A+")
3. Pandas Where with Callable condition¶
You can also pass a callable condition to your where cond parameter. This means you can pass a function to evaluate your value. Here I'll use a lambda function, but you can use a regular function as well.
df.where(lambda x: x<75, "A+")
4. Pandas Where with Callable other¶
Say you wanted to replace your FALSE conditions with a dynamic value. You can also pass a callable into your other parameter
df.where(df<90, lambda x: x**2)
5. Pandas Where with DataFrame other¶
Finally, you can pass a DataFrame into your other parameter to replace your False values. Think of this as selectively copying over values from another DataFrame to your originally DataFrame.
I'll start by making another DataFrame. Notice how only the values that are FALSE in the condition below are the ones that are replaced.
np.random.seed(seed=41)
df2 = pd.DataFrame(data=np.random.randint(0,100,(4,3)),
columns=('Test1', 'Test2', 'Test3'),
index=['Bob', 'Sally', 'Frank', 'Patty']
)
df2
df.where(df<90, df2)
Check out more Pandas functions on our Pandas Page
