On Mon, 19 May 2008 14:38:52 -0700, dc353@hotmail.com <dc353@HOTMAIL.COM> wrote:

>On May 17, 11:24 pm, hs AT dc-sug DOT org ("Howard Schreier)" wrote: >> On Sat, 17 May 2008 19:47:06 -0700, dc...@hotmail.com <dc...@HOTMAIL.COM> wrote: >> >On May 17, 10:37 pm, "dc...@hotmail.com" <dc...@hotmail.com> wrote: >> >> On May 17, 10:16 pm, Andrew H Karp <sfbay0...@aol.com> wrote: >> >> >> > Here is a different approach for you to consider. >> >> >> > data dates; >> >> > do xdate = '01jan2003'd to '01apr2004'd by 28; >> >> > date = intnx('month',(xdate-1),1,'end'); >> >> > output; >> >> > end; >> >> > drop xdate; >> >> > format date mmddyy10.; >> >> > run; >> >> >> > Hope this helps. >> >> >> > Andrew Karp >> >> > Sierra Information Serviceswww.sierrainformation.com >> >> >> > On May 17, 5:57�pm, "dc...@hotmail.com" <dc...@hotmail.com> wrote: >> >> >> > > On May 17, 8:29�pm, "dc...@hotmail.com" <dc...@hotmail.com> wrote: >> >> >> > > > Hi, >> >> >> > > > I'm trying to generate yyyymm dates in a do loop. �The following code >> >> > > > works but doesn't rely on SAS dates. �Is there an easier way using SAS >> >> > > > dates?? >> >> >> > > > DATA _NULL_; >> >> > > > startdate = 200301; >> >> > > > enddate � = 200704; >> >> >> > > > yr = int(enddate/100) - int(startdate/100); >> >> > > > mt = (enddate - int(enddate/100)*100) - (startdate - int(startdate/ >> >> > > > 100)*100); >> >> > > > if mt < 0 then do; >> >> > > > � �yr = yr - 1; >> >> > > > � �mt = mt + 12; >> >> > > > end; >> >> >> > > > cnt = yr*12 + mt; >> >> > > > syr = int(startdate/100); >> >> > > > smt = startdate - syr*100; >> >> >> > > > do i = 0 to cnt; >> >> > > > yr_inc = int((smt+i)/12); >> >> > > > mt_inc = mod(smt+i,12); >> >> > > > if mt_inc = 0 then mt_inc=12; >> >> >> > > > newdate= (syr+yr_inc)*100 + mt_inc; >> >> > > > put newdate; >> >> > > > end; >> >> >> > > > run; >> >> >> > > > What would be nice is something like: >> >> >> > > > do i = startdate to enddate (increment month by 1); >> >> > > > end; >> >> >> > > Ok, here's what I've been able to come up with: �I think this is >> >> > > what's going on: �We take a numeric value and PUT it into a character >> >> > > format, then when take the character value and INPUT it into a date >> >> > > format. >> >> >> > > DATA _NULL_; >> >> > > startdate = 200301; >> >> > > enddate � = 200404; >> >> >> > > startdate=input(put(startdate,6.),yymmn6.); >> >> > > enddate=input(put(enddate,6.),yymmn6.); >> >> > > cnt = intck('month',startdate,enddate); >> >> > > put cnt; >> >> > > do i = 0 to cnt; >> >> > > newdate = intnx('month',startdate,i,'end'); >> >> >> > > format newdate yymmddn8.; >> >> > > put �newdate ; >> >> >> > > end; >> >> >> > > run;- Hide quoted text - >> >> >> > > - Show quoted text - >> >> >> thanks Andrew. Pretty compact. >> >> >I did a little more testing. if I have a longer difference between the >> >start and end date we get duplicate dates every once in awhile - this >> >is happening because of the BY 28. Using the INTCK function avoids >> >the problem. >> >> Try this variation: >> >> data dates; >> do until (date = '01apr2004'd); >> date = ifn( missing(date) >> , '01jan2003'd >> , intnx('month',date,1) >> ); >> output; >> end; >> format date mmddyy10.; >> run; > >Howard, that's pretty cool. do I read this as: if the date is >missing start with 01jan2003 and if the date is not missing intnx it >by 1 month? I've not seen the expression missing(date).

It's a function. See http://support.sas.com/onlinedoc/913/getDoc/en/lrdict.hlp/a000509996.htm

I could have used the NMISS function, or the N function, or a comparison operator.