Добавление продолжительности времени (как int) ко времени (как объект)

Question

У меня есть этот фрейм данных:

     date time_start  duration_seconds
 10-01-18   12:56:52              1382
 10-01-18   12:56:52               164
 10-01-18   23:56:00               250

Типы:

date                datetime64[ns]
time_start                  object
duration_seconds             int64

Я хочу добавить duration_seconds к time_start и создать столбец time_end. Кроме того, для объединения time & date для времени начала и окончания.

Ожидаемый результат:

     date time_start  duration_seconds  time_end time_start_with_date time_end_with_date
 10-01-18   12:56:52              1382  13:19:54    10-01-18 12:56:52  10-01-18 13:19:54
 10-01-18   12:56:52               164  12:59:36    10-01-18 12:56:52  10-01-18 12:59:36
 10-01-18   23:56:00               250  00:00:10    10-01-18 23:56:00  11-01-18 00:00:10

Answer 1

Идея состоит в том, чтобы создать datetime столбец to_datetime и timedeltas to_timedelta и столбцы суммы:

df['time_start'] = pd.to_timedelta(df['time_start'])
df['time_end'] = pd.to_timedelta(df['duration_seconds'], unit='s') + df['time_start']
df['date'] = pd.to_datetime(df['date'], format='%d-%m-%y')

df['time_start_with_date'] = df['date'] + df['time_start']
df['time_end_with_date'] = df['date'] + df['time_end']
print (df)
        date time_start  duration_seconds        time_end  \
0 2018-01-10   12:56:52              1382 0 days 13:19:54   
1 2018-01-10   12:56:52               164 0 days 12:59:36   
2 2018-01-10   23:56:00               250 1 days 00:00:10   

  time_start_with_date  time_end_with_date  
0  2018-01-10 12:56:52 2018-01-10 13:19:54  
1  2018-01-10 12:56:52 2018-01-10 12:59:36  
2  2018-01-10 23:56:00 2018-01-11 00:00:10  

Для пользовательского формата datetimes:

df['time_start'] = pd.to_timedelta(df['time_start'])
df['time_end'] = pd.to_timedelta(df['duration_seconds'], unit='s') + df['time_start']
date = pd.to_datetime(df['date'], format='%d-%m-%y')

df['time_start_with_date'] = (date + df['time_start']).dt.strftime('%d-%m-%y %H:%M:%S')
df['time_end_with_date'] = (date + df['time_end']).dt.strftime('%d-%m-%y %H:%M:%S')

#https://stackoverflow.com/a/51102096/2901002
#solution convert only HH:MM:SS to strings, removed all days
df['time_start'] = df['time_start'].astype(str).astype(str).str[-18:-10]
df['time_end'] = df['time_end'].astype(str).astype(str).str[-18:-10]
print (df)
       date time_start  duration_seconds  time_end time_start_with_date  \
0  10-01-18   12:56:52              1382  13:19:54    10-01-18 12:56:52   
1  10-01-18   12:56:52               164  12:59:36    10-01-18 12:56:52   
2  10-01-18   23:56:00               250  00:00:10    10-01-18 23:56:00   

  time_end_with_date  
0  10-01-18 13:19:54  
1  10-01-18 12:59:36  
2  11-01-18 00:00:10