Find the row associated with maximum date after groupby in Pandas

Question

I have a pandas DataFrame with 3 columns containing a PERSON_ID, MOVING_DATE AND PLACE as follows:

df = pandas.DataFrame(
[[1,datetime.datetime(2018, 1, 1), 'New York'], 
 [1, datetime.datetime(2018, 1, 20), 'Rio de Janeiro'],
 [1, datetime.datetime(2018, 2, 13), 'London'],
 [2, datetime.datetime(2017, 6, 12), 'Seatle'],
 [2, datetime.datetime(2016, 10, 10), 'New Mexico'],
 [3, datetime.datetime(2017, 9, 19), 'Sao Paulo'],
 [3, datetime.datetime(2015, 12, 11), 'Bangladesh']]],
columns=['PERSON ID', 'MOVING DATE', 'PLACE']
)

   PERSON ID MOVING DATE           PLACE
0          1  2018-01-01        New York
1          1  2018-01-20  Rio de Janeiro
2          1  2018-02-13          London
3          2  2017-06-12          Seatle
4          2  2016-10-10      New Mexico
5          3  2017-09-19       Sao Paulo
6          3  2015-12-11      Bangladesh

I would like to find the place where the person is based on its last movement date (MOVEMENT_DATE).

Is it possible to get the result with the groupby method?

So far, I've tried:

df = df.sort_values(['PERSON ID', 'MOVING DATE'])
df.groupby(['PERSON ID', 'MOVING DATE']).agg(
     {'MOVING DATE': max, 'PLACE': 'last'}
)

but it didn't work out. Any help would be appreciated.

Thanks in advance,

Rhenan

Answer 1

A one-liner using DataFrame.groupby and Grouper.last:

df.sort_values('MOVING DATE').groupby('PERSON ID').last()

output:

     MOVING DATE      PLACE
PERSON ID                       
1          2018-02-13     London
2          2017-06-12     Seatle
3          2017-09-19  Sao Paulo

Answer 2

A sort is overkill here, that's O(nlogn) time complexity, when you can do this with loc and idxmax:

df.loc[df.groupby('PERSON ID')['MOVING DATE'].idxmax()]

   PERSON ID MOVING DATE      PLACE
2          1  2018-02-13     London
3          2  2017-06-12     Seatle
5          3  2017-09-19  Sao Paulo