如何用timedelta（python）减去两个时间列？

Question

I'm trying for hours to do a subtraction between this two time columns so I can see how long did it take to the other action happen:

In[1]:aumento_data_separada 
Out[1]: Response_Time      Request_Time(when the client do a request)   
 0   00:56:58.612000    00:46:34.347000
 1   12:00:41.069000    12:00:32.603000
 2   18:05:02.776000    16:39:42.682000
 3   19:27:11.997000    17:33:05.589000 
...

This dataframe has more 392560 rows like these above. I'd like to do the operation in all of them. and creat a new column with the result. I tried reading the docs from timedelta, but I didn't understand it at all. I did this -> type(aumento_data_separada['Response_Time'].iloc[0]) and type(aumento_data_separada['Request_Time'].iloc[0]) to see the type and it returned me a datetime.time . I found out that I can't do operations within them because of the type. So I tried to convert them separately: pd.to_datetime(aumento_data_separada['Request_Time']) but it returns me the following error-> object of type 'datetime.time' has no len() .

Here is the 1st error message and the code I tried -> aumento_data_separada.Response_Time - aumento_data_separada.Request_Time - returning unsupported operand type(s) for -: 'datetime.time' and 'datetime.time'

Answer 1

convert the columns to pd.Timedelta with pd.to_timedelta

df = df.apply(pd.to_timedelta)

df.Response_Time - df.Request_Time

0   00:10:24.265000
1   00:00:08.466000
2   01:25:20.094000
3   01:54:06.408000
dtype: timedelta64[ns]

Answer 2

This is similar to piRSquared's answer, but for some reason this is faster when I test the two methods side-by-side. Not sure why that is.

df['Time_Diff'] = pd.to_datetime(df.Response_Time) - pd.to_datetime(df.Request_Time)