Support for bimonthly/weekly timerules

[JWCornV]

Any change of getting bi-monthly (1st and 15th) of the month, or bi-weekly (every two weeks) supported?

I could probably do this.

[wesm]

I'd be happy to accept a patch for bi-monthly. Bi-weekly is already supported but not in "anchored form":

In [2]: date_range('6/26/2012', periods=10, freq='2W-TUE')
Out[2]: 
<class 'pandas.tseries.index.DatetimeIndex'>
[2012-06-26 00:00:00, ..., 2012-10-30 00:00:00]
Length: 10, Freq: 2W-TUE, Timezone: None

In [3]: list(date_range('6/26/2012', periods=10, freq='2W-TUE'))
Out[3]: 
[<Timestamp: 2012-06-26 00:00:00>,
 <Timestamp: 2012-07-10 00:00:00>,
 <Timestamp: 2012-07-24 00:00:00>,
 <Timestamp: 2012-08-07 00:00:00>,
 <Timestamp: 2012-08-21 00:00:00>,
 <Timestamp: 2012-09-04 00:00:00>,
 <Timestamp: 2012-09-18 00:00:00>,
 <Timestamp: 2012-10-02 00:00:00>,
 <Timestamp: 2012-10-16 00:00:00>,
 <Timestamp: 2012-10-30 00:00:00>]

[dbinetti]

(Edited) The Bi-unit form (multiplied by two) is already supported for any DateOffset object. What you asking for is the Semi-unit form (divided by two), which (AFAIK) is not supported. Having a semi-* form would be useful for many business applications.

[adrienemery]

Any word on if this will be included in an upcoming release? If not I can put together a patch for Semi-Monthly.

I propose:

SM for Semi-Monthly Month End (15th and last day of month)
SMS for Semi-Monthly Month Start (1st and 15th)

[jreback]

This would be a straightforward extension. patches are welcome.