Data Structure
in [SAS]
|
Prev: SAS capabilities for determining covariate coefficients
Next: Call for Paper: International Journal of Mathematics and Computation (IJMC).
From: montura on 13 May 2010 09:43 If you need to maintain one-per-subject integrity for subsequent operations, SCL object arrays are the only option. First you have to declare an object with a three dimensional array: dcl num data [*,*,*]; Then identify the max elements for each dimension. data=makearray(500,4, 1000); 500 subjects, 4 visits, 1000 records max per subject*visit. Load the arrays with SCL GETVARC GETVARN statements. Then you can spin through and slice-and-dice. So maybe the real question is more like "should you use arrays" for that purpose?
From: Al on 17 May 2010 10:13
On May 13, 8:43 am, montura <montura...(a)gmail.com> wrote: > If you need to maintain one-per-subject integrity for subsequent > operations, SCL object arrays are the only option. > > First you have to declare an object with a three dimensional array: > dcl num data [*,*,*]; > > Then identify the max elements for each dimension. > data=makearray(500,4, 1000); 500 subjects, 4 visits, 1000 records max > per subject*visit. > > Load the arrays with SCL GETVARC GETVARN statements. > Then you can spin through and slice-and-dice. > > So maybe the real question is more like "should you use arrays" for > that purpose? All: I was able to get this using arrays ,but what i meant is to mimic the following code using arrays ..i.e to get the vis data values as columns .. and values of res under these columns .... proc transpose data = have out = trans; by pat ; var res; id vis; run ; I appreciate your help .. Thanks Al |