Here is one way to tackle it:

1. sort the data by patno and abs(lbdy) 2. use first.patno to get the record from the sorted data and assign vis=5 3. merge back to original data

All very simple step, you should try yourself.

On Thu, 18 Jun 2009 09:10:16 -0700, Al <ali6058@GMAIL.COM> wrote:

>On Jun 18, 9:23 am, snoopy...@GMAIL.COM (Joe Matise) wrote: >> I recommend a more clear explanation of how you determine vis_num; I have no >> idea why you gave patno #1 a vis_num only for the second row, or any of the >> others really; I suspect if you spend a bit more time considering the logic >> of your request, your answer will follow fairly quickly. >> >> In general, if you want to consider all of the records at once for a >> particular patno, and then retroactively do (something), you can use the >> double DoW loop. For your data you seem to separate the low lb_day from the >> high lb_day, so you'd need an artificial variable to do that (as you need to >> have proper BY group sorting). Something like this probably will work, >> though you note I left a bit where I don't follow your logic so I can't fill >> that bit in. >> >> data have; >> input patno lbtest $ lb_day lbres; >> if lb_day le 7 then sortord=1; >> else sortord=2; >> datalines; >> 1 Creat 4 32 >> 1 Creat 7 12 >> 1 Creat 6 11 >> 1 Creat 2 10 >> 2 Creat 5 30 >> 2 Creat 6 23 >> 2 Creat 7 19 >> 3 Creat 2 16 >> 3 Creat 3 17 >> 3 Creat 4 18 >> 1 Creat 12 32 >> 1 Creat 13 12 >> 1 Creat 14 11 >> 1 Creat 16 10 >> 2 Creat 15 30 >> 2 Creat 16 23 >> 2 Creat 14 19 >> 3 Creat 13 16 >> 3 Creat 14 17 >> 3 Creat 15 18 >> ;;;; >> run; >> >> data want; >> retain vis_num visrec; >> do _n_ = 1 by 1 until (last.patno); >> set have; >> by sortord patno; >> *your logic here to figure out your vis_num; >> if (something) then do; >> visrec = lb_day; >> if sortord = 1 then vis_num = 5; >> else if sortord = 2 then vis_num = 15; >> end; >> end; >> do _n_ = 1 by 1 until (last.patno); >> set have; >> by sortord patno; >> if lb_day ne visrec then call missing(vis_num); *to clear vis_num from >> records other than the desired record; >> output; >> end; >> run; >> >> -Joe >> >> >> >> On Thu, Jun 18, 2009 at 7:34 AM, Al <ali6...@gmail.com> wrote: >> > All, >> >> > I have a dataset as following >> >> > Patno lbtest lb_day lbres vis_num >> > 1 Creat 4 32 >> > 1 Creat 7 12 >> > 1 Creat 6 11 >> > 1 Creat 2 10 >> > 2 Creat 5 30 >> > 2 Creat 6 23 >> > 2 Creat 7 19 >> > 3 Creat 2 16 >> > 3 Creat 3 17 >> > 3 Creat 4 18 >> > 1 Creat 12 32 >> > 1 Creat 13 12 >> > 1 Creat 14 11 >> > 1 Creat 16 10 >> > 2 Creat 15 30 >> > 2 Creat 16 23 >> > 2 Creat 14 19 >> > 3 Creat 13 16 >> > 3 Creat 14 17 >> > 3 Creat 15 18 >> >> > I have to create a variable vis_num based on the value of >> > lb_day.following the below rules >> > for lb_day between 2 and 7 if lb_day = 5 then vis_num should be 5 >> > if >> > lb_day doesnt have a value (i.e 5 in this case)..then i have to pick >> > a >> > value which is closer to 5 ..if equi distant .. i should be able to >> > pick the latter value ..and same with 12 and 15 .. i should be >> > assiging 15. >> >> > The output should look like below.I guess .. i was very crude in >> > explaining the issue. >> >> > Patno lbtest lb_day lbres vis_num >> > 1 Creat 4 32 . >> > 1 Creat 7 12 5lab_day (7,6) equidistant >> > from 5.latter was picked(7) >> > 1 Creat 6 11 . >> > 1 Creat 2 10 . >> > 2 Creat 5 30 5 >> > 2 Creat 6 23 . >> > 2 Creat 7 19 . >> > 3 Creat 2 16 . >> > 3 Creat 3 17 . >> > 3 Creat 4 18 5lab_day(4) closer to 5 ..so >> > it was picked >> > 1 Creat 12 32 . >> > 1 Creat 13 12 . >> > 1 Creat 14 11 . >> > 1 Creat 16 10 15 >> > 2 Creat 14 30 . >> > 2 Creat 15 23 15 >> > 2 Creat 16 19 . >> > 3 Creat 13 16 . >> > 3 Creat 14 17 . >> > 3 Creat 15 18 15 >> >> > Thanks in advance- Hide quoted text - >> >> - Show quoted text - > > >Hi ALL; > > > Here you go again > patno test lbdy > 1 cret 5 > 1 cret 8 > 2 cret 4 > 2 cret 3 > 3 cret 2 > 3 cret 6 > 3 cret 7 > I am supposed to create a variable vis > if lbdy = 5 then vis = 5 > if lbdy doesnt have 5 in it ..i should be able > to pick a row and assign 5 to it based on which one > is closer to 5 , and if equdistant i should be able to pick > the latter .. > > > Here is the output .. > > patno test lbdy vis > 1 cret 5 5 > 1 cret 8 . > 2 cret 4 5 > 2 cret 3 . > 3 cret 4 . > 3 cret 6 5 > 3 cret 7 . > > Thanks in advance and i apolozise for > unclear post, hope i was clear this time