|
On Mon, 18 Sep 2006 14:49:40 -0500, SAS_learner <proccontents@GMAIL.COM> wrote:
>hello guys , I am generating a macro
No macro code in sight. You are using global macro variables, but those are
not the same as macro code.
>like this to used for caliculating
>percentages
There are less roundabout and simpler ways to get there.
>like this
>
> proc freq data=V_AE noprint;
> tables TREAT_R / missing out=freq0(keep=TREAT_R COUNT);
> run;
>
>proc sql noprint;
>create table total as
>select sum(count) as total
>from freq0
>where treat_r ^= ' ' ;
>
>quit;
>data _null_;
> set total;
>call symput("tot1",put(total,3.));
You are using PUT to explicitly convert TOTAL into a 3-byte string.
>call symput ("mftot1","N="|| trim(left(put(total,3.)))!!")");
>run;
>
>to be used as total like this
>
> proc sql noprint;
> create table V_FINAL_cu1 as
>select a.*,a.mi_count/input("&tot1",3.) * 100 as perct ,
>a.mo_count/input("&tot1",3.)
>* 100 as perct1 ,a.se_count/input("&tot1",3.) * 100 as perct2
> from V_FINAL_cu as a ;
You are scanning exactly 3 bytes; all is well.
> quit;
>
> everything is fine when I am doing this way (above ) but when I trying to
>do same thing avoiding call symput (below) the value of &tot is getting
>resolved i.e.
>
>proc sql noprint;
>select sum(count) into : tot1
You are relying on explict numeric-to-character conversion. The result is an
8-byte string with the value right-justfied.
>from freq0
>where treat_r ^= ' ' ;
>%put &tot1;
>quit;
>
>proc sql noprint;
> create table V_FINAL_cu1 as
>select a.*,a.mi_count/input("&tot1",3.) * 100 as perct ,
>a.mo_count/input("&tot1",3.)
>* 100 as perct1 ,a.se_count/input("&tot1",3.) * 100 as perct2
> from V_FINAL_cu as a ;
You are scanning only 3 bytes and thus missing at least part of the value
and likely all of it.
Just the other day, in a different thread, I anticipated such a situation
and cited it as a reason to *not* use the INPUT function in this context. I
would replace
input("&tot1",3.)
with just
&tot1
> quit;
>
> this way &tot1 is not resolving
Yes it is; you're just not scanning it all.
>why ?? and one is there a way I can create
>two macro variable using proc sql similar to what I am doing using call
>symput ??
Sure. A model:
select max(height), min(weight)
into : maxheight, : minweight
from sashelp.class;
>
>thanks for your time and advice.
You might have had an earlier response if you had made it easier to work on
the problem. How? By making it self-contained by including data. It could be
as simple as:
data V_AE;
TREAT_R = 'abc';
run;
data V_FINAL_cu;
retain mi_count 2 mo_count 3 se_count 4;
run;
| 
