help with obs comparisons
in [SAS]
|
Prev: Assigning random numbers
Next: AUTHENTIC DESIGNER HANDBAGS & ACCESSORIES WWW.VOGUELANDE.COM CHANEL LOUIS VUITTON
From: olivesecret on 30 Mar 2010 10:16 I have a data set consisting of obs for many subjects, with about 8 obs for each subject (at 8 time points, one obs/time). Ideally the obs for each subject should be non-increasing over time. If any obs is not in the non-increasing order, I need flag it. For example, obs for subject 1 are: subject res time 1 500 0 1 450 1 1 470 2 1 400 3 1 350 4 1 350 5 1 300 6 1 100 7 Since the 3rd obs >2nd obs, I need flag 3rd. So it looks like I need compare each pair of continuous obs such as 1 vs 2, 2 vs. 3, ..., and so on. How can I do it? Any suggestion is highly appreciated!
From: Paige Miller on 30 Mar 2010 10:47 On Mar 30, 10:16 am, olivesecret <olivesec...(a)gmail.com> wrote: > I have a data set consisting of obs for many subjects, with about 8 > obs for each subject (at 8 time points, one obs/time). Ideally the obs > for each subject should be non-increasing over time. If any obs is not > in the non-increasing order, I need flag it. For example, obs for > subject 1 are: > subject res time > 1 500 0 > 1 450 1 > 1 470 2 > 1 400 3 > 1 350 4 > 1 350 5 > 1 300 6 > 1 100 7 > > Since the 3rd obs >2nd obs, I need flag 3rd. So it looks like I need > compare each pair of continuous obs such as 1 vs 2, 2 vs. 3, ..., and > so on. How can I do it? Any suggestion is highly appreciated! Use the data step LAG function. -- Paige Miller paige\dot\miller \at\ kodak\dot\com
From: Reeza on 30 Mar 2010 16:35
On Mar 30, 7:16 am, olivesecret <olivesec...(a)gmail.com> wrote: > I have a data set consisting of obs for many subjects, with about 8 > obs for each subject (at 8 time points, one obs/time). Ideally the obs > for each subject should be non-increasing over time. If any obs is not > in the non-increasing order, I need flag it. For example, obs for > subject 1 are: > subject res time > 1 500 0 > 1 450 1 > 1 470 2 > 1 400 3 > 1 350 4 > 1 350 5 > 1 300 6 > 1 100 7 > > Since the 3rd obs >2nd obs, I need flag 3rd. So it looks like I need > compare each pair of continuous obs such as 1 vs 2, 2 vs. 3, ..., and > so on. How can I do it? Any suggestion is highly appreciated! Manually in a datastep: Use a by "subject" and the retain to keep the last value in record. Then compare the current value to the prior value. |