Hello SAS users,
I'm running the following code:
data t0;
input Estimate Numord;
datalines;
2447 1
42 2
10 3
-39 4
-84 5
;
run;
data t2;
input beta_3;
datalines;
19
42
10
-42
-76
;
run;
proc sql;
select count(a.beta_3<=b.Estimate)/count(a.beta_3) into :z0bar
from t2 a,
t0 b
where b.numord=3
;
run;
%put _all_;
I am expecting 0.4 for z0bar, but I am getting 1. Why?
select sum(a.beta_3<=b.Estimate)/count(a.beta_3) into :z0bar
What is it your trying to achieve? The reason is, if you look at a table version of your data:
proc sql; create table tmp as select a.beta_3,b.estimate,a.beta_3<=b.Estimate as test,count(a.beta_3<=b.Estimate)/count(a.beta_3) as a format=best. from t2 a, t0 b where b.numord=3 ; run;
You will see that there is 5 rows with a 0 or 1 for each beta_3<=estimate, and the count(beta_3) will always be 1, as there is exactly 1 per row, 0/1=0 and 1/1=1, you get a 1 out.
I'm trying to count the number of rows in b that are <= the value of the variable Estimate at numord=3, as a percentage of all the rows in b.
Thx.
Something like (and this is verbose just to show working out):
create table tmp as select a.*, b.tot, (select count(*) from t2 where beta_3 <= a.estimate) as sub, (calculated sub / b.tot) * 100 as percent from (select * from t0 where numord=3) a left join (select count(*) as tot from t2) b on 1=1; run;
select sum(a.beta_3<=b.Estimate)/count(a.beta_3) into :z0bar
Available on demand!
Missed SAS Innovate Las Vegas? Watch all the action for free! View the keynotes, general sessions and 22 breakouts on demand.
Learn the difference between classical and Bayesian statistical approaches and see a few PROC examples to perform Bayesian analysis in this video.
Find more tutorials on the SAS Users YouTube channel.