CODEWITHSUNDEEP
pythonpandasdatetime
Cleb
Cleb

Reputation: 25997

How to convert datetime series to actual duration in hours?

I have a dataframe like this:

index = ['2018-02-17 00:30:00', '2018-02-17 07:00:00',
'2018-02-17 13:00:00', '2018-02-17 19:00:00',
'2018-02-18 00:00:00', '2018-02-18 07:00:00',
'2018-02-18 10:30:00', '2018-02-18 13:00:00']

df = pd.DataFrame({'col': list(range(len(index)))})
df.index = pd.to_datetime(index)

                     col
2018-02-17 00:30:00    0
2018-02-17 07:00:00    1
2018-02-17 13:00:00    2
2018-02-17 19:00:00    3
2018-02-18 00:00:00    4
2018-02-18 07:00:00    5
2018-02-18 10:30:00    6
2018-02-18 13:00:00    7

and would like to add a column that reflects the actual duration in hours, so my desired outcome looks like this:

                     col  time_range
2018-02-17 00:30:00    0         0.0
2018-02-17 07:00:00    1         6.5
2018-02-17 13:00:00    2        12.5
2018-02-17 19:00:00    3        18.5
2018-02-18 00:00:00    4        23.5
2018-02-18 07:00:00    5        30.5
2018-02-18 10:30:00    6        34.0
2018-02-18 13:00:00    7        36.5

I currently do this as follows:

df['time_range'] = [(ti - df.index[0]).delta / (10 ** 9 * 60 * 60) for ti in df.index]

Is there a smarter (i.e. vectorized/built-in) way of doing this?

Upvotes: 1

Views: 89

Answers (1)

jezrael
jezrael

Reputation: 862661

Use:

df['new'] = (df.index - df.index[0]).total_seconds() / 3600

Or:

df['new'] = (df.index - df.index[0]) / np.timedelta64(1, 'h')

print (df)
                     col   new
2018-02-17 00:30:00    0   0.0
2018-02-17 07:00:00    1   6.5
2018-02-17 13:00:00    2  12.5
2018-02-17 19:00:00    3  18.5
2018-02-18 00:00:00    4  23.5
2018-02-18 07:00:00    5  30.5
2018-02-18 10:30:00    6  34.0
2018-02-18 13:00:00    7  36.5

Upvotes: 2

Related Questions