Hi all,
I can't figure out how in below code the SQL would need to look like so it returns the desired result as done via a data step merge.
I'm looking for something "generic" that would also work for 6 source tables.
I want in the end to generate the code using dictionary tables and I understand that I might end-up with a data step merge with all the renaming "galore" for performance reasons. ...but still really curious how a SQL version would need to look like and I just can't manage to come-up with something simple/a single SQL.
/* create sample data */
data class_A;
set sashelp.class(firstobs=1 obs=4);
call missing(age,weight);
keep name age weight sex;
run;
data class_B;
set sashelp.class(firstobs=3 obs=6);
call missing(age);
keep name age weight sex;
run;
data class_C;
set sashelp.class(firstobs=4 obs=8);
keep name age weight sex;
run;
/* SQL join - not doing the right thing */
proc sql;
create table not_there_yet_want as
select
coalesce(a.name,b.name,c.name) as name
,coalesce(a.age,b.age,c.age) as age
,coalesce(a.weight,b.weight,c.weight) as weight
,coalesce(a.sex,b.sex,c.sex) as sex
from class_A a
full join class_B b
on a.name=b.name
full join class_C c
on a.name=c.name
order by name
;
quit;
title 'SQL not returning desired result';
proc print data=not_there_yet_want;
run;
title;
/* desired result - but looking for a SQL version */
data desired;
if 0 then set sashelp.class(keep=name age weight);
merge
class_a(rename=(age=_a_age weight=_a_weight))
class_b(rename=(age=_b_age weight=_b_weight))
class_c(rename=(age=_c_age weight=_c_weight))
;
by name;
age=coalesce(_a_age,_b_age,_c_age);
weight=coalesce(_a_weight,_b_weight,_c_weight);
drop _:;
run;
title 'Desired result - but not done using SQL';
proc print data=desired;
run;
title;
Thanks,
Patrick
Hi @Patrick
See if this is what you meant:
/* SQL join - not doing the right thing */
proc sql;
create table not_there_yet_want as
select
coalesce(a.name,b.name,c.name) as name
,coalesce(a.age,b.age,c.age) as age
,coalesce(a.weight,b.weight,c.weight) as weight
,coalesce(a.sex,b.sex,c.sex) as sex
from class_A a
full join class_B b
on a.name=b.name
full join class_C c
on coalesce(a.name,b.name)=c.name
order by name
;
quit;
Hi @Patrick
See if this is what you meant:
/* SQL join - not doing the right thing */
proc sql;
create table not_there_yet_want as
select
coalesce(a.name,b.name,c.name) as name
,coalesce(a.age,b.age,c.age) as age
,coalesce(a.weight,b.weight,c.weight) as weight
,coalesce(a.sex,b.sex,c.sex) as sex
from class_A a
full join class_B b
on a.name=b.name
full join class_C c
on coalesce(a.name,b.name)=c.name
order by name
;
quit;
@EyalGonen That did the trick. Thanks!
It's finally time to hack! Remember to visit the SAS Hacker's Hub regularly for news and updates.
Learn how use the CAT functions in SAS to join values from multiple variables into a single value.
Find more tutorials on the SAS Users YouTube channel.
Ready to level-up your skills? Choose your own adventure.