CODEWITHSUNDEEP
pythonpandasnumpy
Daniel
Daniel

Reputation: 351

How to convert pandas time series into numpy array of total elapsed time?

Pandas newbie here. I want to convert a date range time series date_time from pandas:

import pandas as pd
import numpy as np
In [1]: date_time = pd.date_range('2017/10/27','2017/10/29',freq='12H')

into a numpy array t containing the elapsed time (for instance, in hours), such that:

In [2]: t
Out[2]: array([0,12,24,36,48])

What would be the most direct way to do this?

I need this to pass the numpy array t to odeint routines.

Upvotes: 1

Views: 3100

Answers (2)

cs95
cs95

Reputation: 402493

You could also convert to a pd.Series and call dt.total_seconds.

x = pd.Series(date_time)
y = (x - x[0]).dt.total_seconds().div(60 * 60).values
y

array([  0.,  12.,  24.,  36.,  48.])

Upvotes: 1

unutbu
unutbu

Reputation: 879511

You can divide Timedeltas by another Timedelta to convert them to a desired frequency unit (such as hours). Use the .values attribute to convert the result to a NumPy array: 

In [37]: ((date_time - date_time[0]) / pd.Timedelta('1H')).values
Out[37]: array([  0.,  12.,  24.,  36.,  48.])

Upvotes: 1

Related Questions