96

I have a pandas column of Timestamp data

In [27]: train["Original_Quote_Date"][6] 
Out[27]: Timestamp('2013-12-25 00:00:00')

How can check equivalence of these objects to datetime.date objects of the type

datetime.date(2013, 12, 25)
kilojoules
  • 9,768
  • 18
  • 77
  • 149
  • 3
    See also the related question for *datetime*: http://stackoverflow.com/questions/13703720/converting-between-datetime-timestamp-and-datetime64 – Andy Hayden Dec 20 '15 at 23:09

9 Answers9

117

Use the .date method:

In [11]: t = pd.Timestamp('2013-12-25 00:00:00')

In [12]: t.date()
Out[12]: datetime.date(2013, 12, 25)

In [13]: t.date() == datetime.date(2013, 12, 25)
Out[13]: True

To compare against a DatetimeIndex (i.e. an array of Timestamps), you'll want to do it the other way around:

In [21]: pd.Timestamp(datetime.date(2013, 12, 25))
Out[21]: Timestamp('2013-12-25 00:00:00')

In [22]: ts = pd.DatetimeIndex([t])

In [23]: ts == pd.Timestamp(datetime.date(2013, 12, 25))
Out[23]: array([ True], dtype=bool)
Andy Hayden
  • 359,921
  • 101
  • 625
  • 535
  • 2
    For an entire column or series, just use this in conjunction with an apply method and lambda. For example, if t is a series of timestamps: t.apply(lambda x: x.date()) – Den Thap Dec 02 '19 at 19:41
  • 4
    It is worth to mention that the time part is _lost_ and only date part is kept. For those who need to keep time, use `.to_pydatetime()` as mentioned by Xavier Ho. – Mincong Huang Feb 23 '20 at 08:27
  • There is a corresponding `.time()` method that drops the date and returns just the `datetime.time` component – BallpointBen Mar 01 '21 at 16:29
32

As of pandas 0.20.3, use .to_pydatetime() to convert any pandas.DateTimeIndex instances to Python datetime.datetime.

Xavier Ho
  • 17,011
  • 9
  • 48
  • 52
  • 7
    Worth noting that for large DatetimeIndexs this can be slow / lot of memory. This is because a DatetimeIndex is basically just a light wrapper around an array of int64s, whilst an array of python datetimes is an array of fully-fledged python objects/not compactly laid out. – Andy Hayden Mar 18 '18 at 19:45
7

Assume time column is in timestamp integer msec format

1 day = 86400000 ms

Here you go:

day_divider = 86400000

df['time'] = df['time'].values.astype(dtype='datetime64[ms]') # for msec format

df['time'] = (df['time']/day_divider).values.astype(dtype='datetime64[D]') # for day format
Alperen Sözer
  • 371
  • 3
  • 7
6

You can convert a datetime.date object into a pandas Timestamp like this:

#!/usr/bin/env python3
# coding: utf-8

import pandas as pd
import datetime

# create a datetime data object
d_time = datetime.date(2010, 11, 12)

# create a pandas Timestamp object
t_stamp = pd.to_datetime('2010/11/12')

# cast `datetime_timestamp` as Timestamp object and compare
d_time2t_stamp = pd.to_datetime(d_time)

# print to double check
print(d_time)
print(t_stamp)
print(d_time2t_stamp)

# since the conversion succeds this prints `True`
print(d_time2t_stamp == t_stamp)
albert
  • 8,027
  • 10
  • 48
  • 84
5

I've used the way recommended by Filomeno Gonzalez, albeit with a slight twist:

 data['date'] = data['date'].apply(lambda x: x.date())
maynouf
  • 103
  • 1
  • 6
5
from datetime import datetime

time = datetime.fromtimestamp(1676266245263 / 1000)

Example Output : 2023-02-13 05:30:45.263000

Suraj Rao
  • 29,388
  • 11
  • 94
  • 103
Shiwani
  • 51
  • 1
  • 3
3

So, got this from an IBM coursera tutorial.

data['date'] = data['TimeStamp'].apply(lambda d: datetime.date.fromtimestamp(d))
Ali karimi
  • 371
  • 3
  • 10
Filomeno Gonzalez
  • 39
  • 1
0

If I have a pandas DataFrame with timestamp column (1546300800000, 1546301100000, 1546301400000, 1546301700000, 1546302000000) and I want to convert this into date time format

import datetime

df['date'] = df['date'].apply(lambda x: datetime.datetime.fromtimestamp(x/1000.0))

This will return a column with the format 2019-01-01 00:00:00, 2019-01-01 00:05:00, 2019-01-01 00:10:00, 2019-01-01 00:15:00, 2019-01-01 00:20:00...etc

Dividing by 1000 to convert from ms to s as explained here

Hom Bahrani
  • 3,022
  • 27
  • 25
0

In my case, I had the epochs as a normal "column" (i.e. it wasn't a TimeSeries).

I converted it as follows:

import pandas as pd
df['Timestamp'] = pd.to_datetime(df['Timestamp'] / 1000, unit='s')

Time was in milliseconds, so I needed to divide by a thousand. I set the unit='s' to mention that the Timestamp is in seconds (you can, however, use unit='ms' if you don't want to divide by 1000, or depending on your original Timestamp unit).

A sample result

1673885284000 becomes 2023-01-16 4:08:04
Yahya
  • 13,349
  • 6
  • 30
  • 42