How to turn a series that contains pandas.datetime.timedelta to floats e.g. hours or days?

Question

I want to plot some data that contains timedeltas. However, the numpy plot() and errorbar() return:

TypeError: float() argument must be a string or a number, not 'Timedelta'

The question is how can I efficiently convert the timedeltas to a float or integer number. E.g. when I use minutes as base:

0 days 01:30:00 -> 90

or hours

0 days 01:30:00 -> 1.5

Similar to Series.dt.days which would return 0 in this case.

Sören · Answer

One way to do this would be to use Series.dt.seconds and ´Series.dt.days´ and  multiply with a factor for the desired unit:

(Series.dt.seconds/3600) + (Series.dt.days*24)  # for values with [hours]

pqnet · Answer

You can divide by the timedelta you want to use as unit:
totalDays = my_timedelta = pd.TimeDelta('1D')

2 Answers