Wensui,

I think you need more explanation of your rule. The new rule seems to conflict with the rule in the original post. Unless you have a clear defined rule, I'm afraid that people here will not be able to help you.

Kind regards,

Ya Huang

On Sat, 19 Mar 2005 09:45:19 -0500, Wensui Liu <liuwensui@GMAIL.COM> wrote:

>Thank you all for your reply. Maybe I didn't explain clearly. > >What I want is: >if weeks = 1, the sum = supply >if weeks = 2, then sum = sum(supply, lag(supply)) >if weeks = 3, then sum = sum(supply, lag(supply), lag2(supply)) > >Yes, there is no pattern at all, really depending on the value of >weeks. Any idea? > > >On Sat, 19 Mar 2005 08:44:09 -0500, Arthur Tabachneck ><art297@netscape.net> wrote: >> Wensui, >> >> Hopefully, your example was flawed and the following is what you are >> looking for: >> >> data one; >> input supply weeks; >> cards; >> 12 3 >> 22 1 >> 1 2 >> 3 1 >> 2 3 >> ; >> run; >> proc sort data=one; >> by weeks; >> run; >> proc means sum; >> var supply; >> by weeks; >> run; >> >> Which will result in week1=25, week2=1 and week3=14. >> >> Art >> ------ >> On Sat, 19 Mar 2005 01:22:15 -0500, Wensui Liu <liuwensui@GMAIL.COM> wrote: >> >> >Dear All, >> > >> >I have a problem on how to calculate the sum. >> > >> >supply weeks >> > 12 3 >> > 22 1 >> > 1 2 >> > 3 1 >> > 2 3 >> > >> >If weeks = 3, I want sum(12, 22, 1). If weeks = 1, then sum(22). If >> >weeks = 2, then sum(1, 3) and so on.... >> > >> >I have a big dataset with similar data like this and the weeks ranging >> >from 1 to 8. I don't know how to calcuate this summation based on >> >different values of weeks. >> > >> >Thank you so much for your help. >> > > >-- >WenSui Liu, MS MA >Senior Decision Support Analyst >Division of Health Policy and Clinical Effectiveness >Cincinnati Children Hospital Medical Center