Efficient use of numpy select

I have some data as follows. I’m trying to compute the values in the Time bw column (it should be 0 in the 4th row). Whenever the Location moves to a new one, like a to b, I want the Time bw to restart from 0. I’m trying to use ne ,select and diff()

+----------+---------------------+----------+ | Location |         Date        | Time bw  | +----------+---------------------+----------+ | a        | 2018-06-26 00:00:00 |        0 | | a        | 2018-06-26 00:00:00 |        0 | | a        | 2018-06-26 00:00:00 |        0 | | b        | 2018-08-03 00:00:00 |       38 | | b        | 2018-08-03 00:00:00 |        0 | | b        | 2018-08-04 00:00:00 |        1 | | b        | 2018-08-04 00:00:00 |        0 | | b        | 2018-08-04 00:00:00 |        0 | | b        | 2018-08-04 00:00:00 |        0 | | b        | 2018-08-04 00:00:00 |        0 | | b        | 2018-08-04 00:00:00 |        0 | | b        | 2018-08-05 00:00:00 |        1 | | b        | 2018-08-08 00:00:00 |        3 | | b        | 2018-08-08 00:00:00 |        0 | | b        | 2018-08-08 00:00:00 |        0 | | b        | 2018-08-08 00:00:00 |        0 | | b        | 2018-08-08 00:00:00 |        0 | | c        | 2018-08-14 00:00:00 |        6 | | c        | 2018-08-14 00:00:00 |        0 | | c        | 2018-08-14 00:00:00 |        0 | 

+———-+———————+———-+