Convert datetime series to numeric timestamp

Question

I know that it's possible to convert a single datetime observation to Unix timestamp by:

import datetime as dt
int(dt.datetime.now().timestamp())

How could I do this conversion with a Series like this:

import pandas as pd
x = dt.datetime.now()
y = x + dt.timedelta(seconds=300)
z = pd.Series([x, y])

z:

0   2019-03-22 13:53:42.671575
1   2019-03-22 13:58:42.671575
dtype: datetime64[ns]

Use `print (z.astype(np.int64) // 10**9)` – jezrael Mar 22 '19 at 13:33 — jezrael, Mar 22 '19 at 13:33

Manuel · Answer 1 · 2019-03-26T16:12:45.830

0

z = pd.Series([int(x.timestamp()), int(y.timestamp())])

z

0    1553259897
1    1553260197
dtype: int64

.

z = []
for i,s in enumerate(array):
    z[i] = pd.Series([int(s[0].timestamp()), int(s[1].timestamp())])

edited Mar 26 '19 at 16:12

answered Mar 22 '19 at 13:06

Manuel

I have a Series with up to 20,000 time points, is there a more efficient way to accomplish this? – José Vallejo Mar 22 '19 at 13:10
Sure, make a loop and you will have it: z = [] For i,s in enumerate(Series): ----z[i] = pd.Series([int(s[0].timestamp()), int(s[1].timestamp())]) – Manuel Mar 26 '19 at 16:06
I edited my comment above to show the loop. – Manuel Mar 26 '19 at 16:13

1 Answers1