Pandas - operations on groups using transform

Question

Here is my example:

import pandas as pd
import numpy as np

df = pd.DataFrame({'A A': ['one', 'one', 'two', 'two', 'one'] ,
                   'B': ['Ar', 'Br', 'Cr', 'Ar','Ar'] ,
                   'C': ['12/15/2011', '11/11/2001', '08/30/2015', '07/3/1999','03/03/2000' ],
                      'D':[1,7,3,4,5]})

df['C'] = pd.to_datetime(df['C'])

def date_test(x):
    key_date = pd.Timestamp(np.datetime64('2015-08-13'))
    end_date = pd.Timestamp(np.datetime64('2016-10-10'))
    result = False

    for i in x.index:
        if key_date < x[i] < end_date:
            result = True

    return result

def int_test(x):
    result = False
    for i in x.index:
        if 1 < x[i] < 9:
            result = True

    return result

Now I am grouping by column B and transforming column C and D

The following code producess column of ones.

df.groupby(['B'])['D'].transform(int_test)

And the following code produces column of dates

df.groupby(['B'])['C'].transform(date_test)

I would expect them both to produce collection of ones and zeros and not dates. My goal is to get collection of ones and zeros. Any thoughts?

Update: My main goal is to understand how transform works.

Answer 1

For type consistency with subsequent operations your can do with the results of a transform call, that function tries to cast the resulting Series into the dtype of the selected data it works against. The function source code has this dtype cast explicitly done.

Your boolean data can be turned into dates, thus you obtain a datetime series. Explicitly cast to int to get the expected type:

df.groupby(['B'])['C'].transform(date_test).astype('int64')