how to program like this?

[Linlin]

Thank you Mike and Astounding! I am happy because my code is the shortest:smileysilly::smileylaugh:!

Linlin

[art297]

Linlin,

Your comment reminds me of a smallest program challenge, on SAS-L, some years ago.  Take a look at:

Search results -- SAS-L</title><style type="text/css"><!--BODY { font-family: "Comic Sans MS",arial,...

[Linlin]

Art,

I have looked at the link. Too bad you did not win.

[art297]

Linlin: ah, but I did (in a sense).  That, and about another gazillion similar posts, got me into the SAS-L Hall of Fame (see: http://www.sascommunity.org/wiki/SAS-L_BOF )

[Ksharp]

For your question, I would like to use two skill: self-merge or SQL .

I prefer to SQL , because it might be faster when your table is large. Just a guess .

data have;
input x;
datalines;
1
2
3
4
5
6
7
8
9
10
;
run;
/*self-merge skill*/
data want;
 merge have have(firstobs=2 rename=(x=x1)) have(firstobs=3 rename=(x=x2)) have(firstobs=4 rename=(x=x3)) have(firstobs=5 rename=(x=x4))  have(firstobs=6 rename=(x=x5));
 avg=mean(of x1-x5);
 drop x1-x5;
run;


/*SQL skill*/
data have;
 set have;
 n=_n_;
run;
proc sql;
 create table want(drop=n) as
  select *,(select mean(x) from have where n between a.n+1 and a.n+5) as avg
   from have as a;
quit;

Ksharp

[DanielSantos]

And another way to do it, no arrays, just lag.

data temp (keep=MEAN5_X);

set have;

retain MEAN5_X .;

MEAN5_X=sum(MEAN5_X,X,-lag5(X));

if _N_ gt 5;

run;

data want;

merge have temp;

MEAN5_X=MEAN5_X/5;

run;

Cheers from Portugal.

Daniel Santos @ www.cgd.pt

[TomiKong]

Thank you all for providing so many ideas. Thanks.