converting a pandas date to week number

Question

I would like to extract a week number from data in a pandas dataframe.

The date format is datetime64[ns]

I have normalized the date to remove the time from it

df['Date'] = df['Date'].apply(pd.datetools.normalize_date)

so the date now looks like - 2015-06-17 in the data frame column

and now I like to convert that to a week number.

Thanks in advance

score 126 · Accepted Answer · answered Jul 02 '15 at 09:53

126

Just access the dt week attribute:

In [286]:
df['Date'].dt.week

Out[286]:
0    25
dtype: int64

In [287]:
df['Week_Number'] = df['Date'].dt.week
df

Out[287]:
        Date  Week_Number
0 2015-06-17           25

answered Jul 02 '15 at 09:53

EdChum

339,461
188
752
538

3

Note that the `week` property of the `Timestamp` object is ISO (8601) week of the year: https://github.com/pandas-dev/pandas/blob/bc9d329ba83795845be6aa455178e2f8d753542b/pandas/tests/scalar/timestamp/test_timestamp.py#L156 – tozCSS Jan 22 '20 at 17:48
11

Now this solution raises : FutureWarning: Series.dt.weekofyear and Series.dt.week have been deprecated. Please use Series.dt.isocalendar().week instead. – Adept Oct 19 '20 at 13:58

Axel · Answer 2 · 2018-08-20T09:20:22.707

55

Here is another possibility using strftime. strftime.org is a good resource.

df['Week_Number'] = df['Date'].dt.strftime('%U')

'%U' represents the week number of the year (Sunday as the first day of the week) as a zero padded decimal number. All days in a new year preceding the first Sunday are considered to be in week 0.

If you have dates from multiple years, I recommend creating a Year-Week combination

df['Year-Week'] = df['Date'].dt.strftime('%Y-%U')

edited Aug 20 '18 at 09:20

answered Aug 09 '17 at 14:24

Axel

2,103
2
17
27

4

Best solution for me, since now, the accepted solution raises : FutureWarning: Series.dt.weekofyear and Series.dt.week have been deprecated. Please use Series.dt.isocalendar().week instead. – Adept Oct 19 '20 at 13:57

Sokolokki · Answer 3 · 2019-06-21T11:37:55.470

14

Pandas has its .dayofyear and .weekofyear functionality, which can be applied straight away to the output of pandas.to_datetime(df['column_name']), giving type "Timestamp" as the output.

import pandas as pd
df['formatted_date'] = pd.to_datetime(df['datetime'])
df['day_of_year'] = df.formatted_date.apply(lambda x: x.dayofyear)
df['week_of_year'] = df.formatted_date.apply(lambda x: x.weekofyear)

edited Jun 21 '19 at 11:37

answered Mar 18 '19 at 12:41

Sokolokki

733
7
19

Klaus Smit · Answer 4 · 2019-09-07T21:33:39.317

3

from datetime import date
df_date = pd.DataFrame([date.today()],columns  = ['today'])
print(df_date)
#### Print Output ####
#        today
#0  2019-09-07
df_date['weeknum'] = df_date.today.apply(lambda x:x.isocalendar()[1])
print(df_date)
#### Print Output ####
#        today  weeknum
#0  2019-09-07       36

edited Sep 07 '19 at 21:33

answered Sep 06 '19 at 15:52

Klaus Smit

31
2

7

To improve this answer consider adding some text to support how this solves the problem. – haldo Sep 06 '19 at 16:00

score 2 · Answer 5 · answered Jun 24 '21 at 13:05

Update to this answer
In my current python version (3.7, May 2021). The syntax df['Date'].dt.week is printing the following warning: FutureWarning: weekofyear and week have been deprecated, please use DatetimeIndex.isocalendar().week instead The way to use DatetimeIndex would be: df['week_number'] = pd.DatetimeIndex(df.index).isocalendar().week
Here a small demonstration of its use to return a Series

# Input
time_idx = pd.date_range('2022-01-01', periods=4, freq='H').tz_localize('UTC')
values = [9 , 8, 7, 6]

df1 = pd.DataFrame(data = values, index=time_idx, columns=['vals'])
# FutureWarning: weekofyear and week have been deprecated
df1['week_number'] = df1.index.week 

# Using DatetimeIndex.isocalendar().week instead
df2 = pd.DataFrame(data = values, index=time_idx, columns=['vals']) 
# Does not throws a warning
df2['week_number'] = pd.DatetimeIndex(df2.index).isocalendar().week 

print(df2)

score 0 · Answer 6 · answered Oct 22 '21 at 15:23

In case of pandas:

import random
import pandas as pd

desired_length = 100
desired_frequency="20D" # XXXM: XXX months, "XXXD":XXX days, XXXMin: XXX minutes etc.

index = pd.date_range('2020-01-01', periods=desired_length, freq=desired_frequency)
data = [random.random() for _ in range(len(index))]

df = pd.DataFrame(data=data, index=index, columns=['DATA'])
df[df.index.isocalendar().keys()] = df.index.isocalendar()

converting a pandas date to week number

6 Answers6

Linked