or swap the PROC TRANSPOSE & DATA STEP: *******************************************; data have; array _a[*] a b c d e f g h i (1:9); do _n_ = 1 to 7; output; end; run;

data out; set have; array _a{*} _numeric_; do j=1 to dim(_a); name=cats(vname(_a[j]), _n_); obs=_a[j]; output; end; keep name obs; run;

proc sql; create view outv as select * from out order by name; quit; proc transpose data=outv out=out2 name=name; id name; run;

On Thu, 9 Sep 2010 12:30:43 -0500, Data _null_; <iebupdte@GMAIL.COM> wrote:

>I want to join in too! > >This should work for a resonable number of OBS and VARs. >No counting required. > >data have; > array _a[*] a b c d e f g h i (1:9); > do _n_ = 1 to 7; > output; > end; > run; >proc transpose out=id name=id; >proc transpose out=tall; by id; >data tall; > length _name_ $32; > set tall; > _name_ = tranwrd(_name_,'COL',trim(ID)); > run; >proc transpose out=arrayed(drop=_name_); > run; >proc print; > run; > > > >On 9/9/10, Randall Powers <powers_r@bls.gov> wrote: >> Hello All. >> >> I have this dataset: >> >> obs a b c d e f g h i >> 1 >> 2 >> 3 >> 4 >> 5 >> 6 >> 7 >> >> >> Hence, nine variables, seven observations. >> >> What I'd like to end up with is a dataset with one observation containing >> nine arrays with the previous observation number as the array index. >> Hence,7*9=63 variables: >> >> obs a1 a2 a3 a4 a5 a6 a7 b1 b2 b3....b7.....i7 >> 1 >> >> How do I do this? >> >> Thanks! >>