How to Boolean index a pandas dataframe with a function

Question

Is there a way to use a methods for Boolean indexing in pandas DataFrame?

For example:

import pandas


def filter_func(v) -> bool:
    return v == 'asd'


def main():
    df_test = pandas.DataFrame(
        [
            ['sd'], ['asd'], ['sdf']
        ],
        columns=["col-a"]
    )
    #### ERROR: This next line calls filter_func with all contents of column 'col-a'
    result = df_test[df_test['col-a'] == filter_func(df_test['col-a'])]


if __name__ == '__main__':
    main()

In the example above I want to keep only those values for which filter_func will return True. And a result should contain dataframe with single row, but instead I'm getting empty dataframe.

I understand that instead of executing filter_func for each row it is executed only once.

Is there a way to call it for each row?

Should I use apply or map for Series in this case?

Or is there any other way?

Trenton McKinney · Accepted Answer · 2020-10-06 14:41:10Z

df_test['col-a'] is being filtered by the function, so only [filter_func(df_test['col-a'])] is needed, not [df_test['col-a'] == filter_func(df_test['col-a'])].
pandas: Boolean Indexing

import pandas as pd
import numpy as np
import random

# sample data
np.random.seed(365)
random.seed(365)
rows = 1100
data = {'a': np.random.randint(10, size=(rows)),
        'groups': [random.choice(['1-5', '6-25', '26-100', '100-500', '500-1000', '>1000']) for _ in range(rows)],
        'treatment': [random.choice(['Yes', 'No']) for _ in range(rows)],
        'date': pd.bdate_range(datetime.today(), freq='h', periods=rows).tolist()}
df = pd.DataFrame(data)

   a  groups treatment                date
0  2   >1000       Yes 2020-10-06 00:00:00
1  4  26-100        No 2020-10-06 01:00:00
2  1   >1000       Yes 2020-10-06 02:00:00
3  5    6-25       Yes 2020-10-06 03:00:00
4  2  26-100        No 2020-10-06 04:00:00

# filter function
def filter_func(v) -> bool:
    return v == '26-100'


# call function
filtered = df[filter_func(df.groups)]

# display(filtered)
    a  groups treatment                date
1   4  26-100        No 2020-10-06 01:00:00
4   2  26-100        No 2020-10-06 04:00:00
21  2  26-100       Yes 2020-10-06 21:00:00
24  9  26-100       Yes 2020-10-07 00:00:00
32  5  26-100        No 2020-10-07 08:00:00

Collectives™ on Stack Overflow

How to Boolean index a pandas dataframe with a function

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related