I'd like some help understanding this behavior:

import pandas as pd
pd.date_range("2016-09-01", "2006-03-01", freq="-6MS", inclusive="left")

This returns:

DatetimeIndex(['2016-09-01', '2016-03-01', '2015-09-01', '2015-03-01',
               '2014-09-01', '2014-03-01', '2013-09-01', '2013-03-01',
               '2012-09-01', '2012-03-01', '2011-09-01', '2011-03-01',
               '2010-09-01', '2010-03-01', '2009-09-01', '2009-03-01',
               '2008-09-01', '2008-03-01', '2007-09-01', '2007-03-01',
               '2006-09-01'],
              dtype='datetime64[ns]', freq='-6MS')

Note that here '2006-03-01' is missing.

When I move the end date forward to 2006-03-02... 2006-03-01 now IS included:

import pandas as pd
pd.date_range("2016-09-01", "2006-03-02", freq="-6MS", inclusive="left")

Returns:

DatetimeIndex(['2016-09-01', '2016-03-01', '2015-09-01', '2015-03-01',
               '2014-09-01', '2014-03-01', '2013-09-01', '2013-03-01',
               '2012-09-01', '2012-03-01', '2011-09-01', '2011-03-01',
               '2010-09-01', '2010-03-01', '2009-09-01', '2009-03-01',
               '2008-09-01', '2008-03-01', '2007-09-01', '2007-03-01',
               '2006-09-01', '2006-03-01'],
              dtype='datetime64[ns]', freq='-6MS')

I expected 2006-03-01 to be excluded and the result to be the same result in both cases, why is this happening? It's counting backwards from 2016-09-01 by 6 month intervals, so it shouldn't include 2006-03-01 when the last date is set to a value greater than that (e.g. 2006-03-02), right?

I'd like some help understanding this behavior:

import pandas as pd
pd.date_range("2016-09-01", "2006-03-01", freq="-6MS", inclusive="left")

This returns:

DatetimeIndex(['2016-09-01', '2016-03-01', '2015-09-01', '2015-03-01',
               '2014-09-01', '2014-03-01', '2013-09-01', '2013-03-01',
               '2012-09-01', '2012-03-01', '2011-09-01', '2011-03-01',
               '2010-09-01', '2010-03-01', '2009-09-01', '2009-03-01',
               '2008-09-01', '2008-03-01', '2007-09-01', '2007-03-01',
               '2006-09-01'],
              dtype='datetime64[ns]', freq='-6MS')

Note that here '2006-03-01' is missing.

When I move the end date forward to 2006-03-02... 2006-03-01 now IS included:

import pandas as pd
pd.date_range("2016-09-01", "2006-03-02", freq="-6MS", inclusive="left")

Returns:

DatetimeIndex(['2016-09-01', '2016-03-01', '2015-09-01', '2015-03-01',
               '2014-09-01', '2014-03-01', '2013-09-01', '2013-03-01',
               '2012-09-01', '2012-03-01', '2011-09-01', '2011-03-01',
               '2010-09-01', '2010-03-01', '2009-09-01', '2009-03-01',
               '2008-09-01', '2008-03-01', '2007-09-01', '2007-03-01',
               '2006-09-01', '2006-03-01'],
              dtype='datetime64[ns]', freq='-6MS')

I expected 2006-03-01 to be excluded and the result to be the same result in both cases, why is this happening? It's counting backwards from 2016-09-01 by 6 month intervals, so it shouldn't include 2006-03-01 when the last date is set to a value greater than that (e.g. 2006-03-02), right?

• python
• pandas

• Please make sure the title matches the body; I see I didn't capture accurately your intent in the staging area :). – AD7six Commented Apr 1 at 8:40

2 Answers 2

In this line

pd.date_range("2016-09-01", "2006-03-01", freq="-6MS", inclusive="left") date 2006-03-01

the date 2006-03-01 is excluded because inclusive='left' omits the end date if it falls on the boundary, as stated on https://pandas.pydata./docs/reference/api/pandas.date_range.html.

When the previous date is 2006-09-01, the ending date, after adding a 6-month interval, is 2006-03-01. It lands exactly on that date "2006-03-01".

However, dates within the same landing month are included as long as the end date is later than the generated date. Specifically, 2006-03-01 will be included if the end date falls between 2006-03-02 and 2006-03-31.

This code will inlcude 2006-03-01 :

import pandas as pd
dates = pd.date_range("2016-09-01", "2006-03-31", freq="-6MS", inclusive="left")
print(dates)

However, if the ending date is in the fourth month, the date 2006-03-01 will not be included:

import pandas as pd
dates = pd.date_range("2016-09-01", "2006-04-01", freq="-6MS", inclusive="left")
print(dates)

To explicitly include 2006-03-01, you may use inclusive="both", which ensures that both the start and end dates are included. Alternatively, omitting the inclusive parameter defaults to "both", achieving the same result.

It's because you're setting the inclusive argument to left. Set it to both or don't set it as both is the default.

pd.date_range("2016-09-01", "2006-03-01", freq="-6MS")

DatetimeIndex(['2016-09-01', '2016-03-01', '2015-09-01', '2015-03-01',
               '2014-09-01', '2014-03-01', '2013-09-01', '2013-03-01',
               '2012-09-01', '2012-03-01', '2011-09-01', '2011-03-01',
               '2010-09-01', '2010-03-01', '2009-09-01', '2009-03-01',
               '2008-09-01', '2008-03-01', '2007-09-01', '2007-03-01',
               '2006-09-01', '2006-03-01'],
              dtype='datetime64[ns]', freq='-6MS')