How can I take the last row per user in a data frame

Asked Mar 22 '21 at 13:22

Active Mar 22 '21 at 13:55

Viewed 19 times

How are you? How can I take the last row per user in a data frame (each user has many services, the dataframe is biger but i gonna do a small example) example:

pd.DataFrame({"user":['1','1','1','1','1','2','2','3','4'],
             "service":['a','b','nan','c','b','a','c','a','a'],
            "date":['2020-01-13','2020-01-17','2020-02-18','2020-02-18','2020-03-22','2020-01-13','2020-02-13','2020-01-13','nan']})

I only want the last row per user (because it is the near date), my idea in the future is to create a new column name "service history" and put the array there ... but the first one I only need to take the last raw per user , as below output:

pd.DataFrame({"user":['1','2','3','4'],
         "service":['b','c','a','a'],
        "date":['2020-03-22','2020-02-13','2020-01-13','nan']})

edited Mar 22 '21 at 13:55

asked Mar 22 '21 at 13:22

emanuel lemos

Use `df['date'] = pd.to_datetime(df['date']) df = df.loc[df.groupby('user')['date'].idxmax()]` – jezrael Mar 22 '21 at 13:29
1

@jezrael it's not working because date is a datetime (and not the 100% of data is datetime, some values arer NaN). So Dont return the last row – emanuel lemos Mar 22 '21 at 13:50
Ok, can you change data for see how solution failed? – jezrael Mar 22 '21 at 13:51
So need `df = df.sort_values(['user','date'], ascending=[True, False]).drop_duplicates('user')` ? – jezrael Mar 22 '21 at 13:57

How can I take the last row per user in a data frame

0 Answers0