topic proc sql-Join in SAS Programming

proc sql-Join

Ronein — Wed, 25 Jul 2018 09:19:43 GMT

Hello

Please find following code in proc sql.

I want to ask please what type of join is done here?

Inner join? full join? left join? right join?

Why If I write - on survey.Answer=survey2.Answer I receive an error?

and it only works with - where survey.Answer=survey2.Answer ?

data survey;
    input State $ Answer $ @@;
    datalines;
NY YES NY YES NY YES NY YES NY YES NY YES NY NO  NY NO  NY NO  NC YES 
NC YES NC YES NC YES NC YES NC YES NC YES NC YES NC YES NC YES NC YES
NC YES NC YES NC YES NC YES NC YES NC YES NC YES NC YES NC YES NC NO
NC NO  NC NO  NC NO  NC NO  NC NO  NC NO  NC NO  NC NO  NC NO  NC NO
NC NO  NC NO  NC NO  NC NO  NC NO  NC NO  NC NO  NC NO  NC NO  NC NO
NC NO  NC NO  NC NO  PA YES PA YES PA YES PA YES PA YES PA YES PA YES
PA YES PA YES PA NO  PA NO  PA NO  PA NO  PA NO  PA NO  PA NO  PA NO  
PA NO  PA NO  PA NO  PA NO  PA NO  PA NO  PA NO  PA NO  PA NO  PA NO
VA YES VA YES VA YES VA YES VA YES VA YES VA YES VA YES VA YES VA YES
VA YES VA YES VA YES VA YES VA YES VA YES VA YES VA YES VA YES VA NO
VA NO  VA NO  VA NO  VA NO  VA NO  VA NO  VA NO  VA NO  VA NO  VA NO
VA NO  VA NO  VA NO  VA NO  VA NO  VA NO
;
Run;

proc sql;
   title1 'Survey Responses';
   select survey.Answer, survey.State,
                  count(*) as Count,
                  calculated Count/Subtotal as Percent format=percent8.2
   from survey,
        (select Answer, 
                 count(*) as Subtotal
                 from survey
                  group by Answer) as survey2
   where survey.Answer=survey2.Answer
   group by survey.Answer, survey.State;
quit;

Re: proc sql-Join

RW9 — Wed, 25 Jul 2018 09:49:31 GMT

Interesting question, never really thought about parallels. The on keyword will not work as no join is specified. The join is based on the where clause, so only data where answer is in both datasets would be output, so inner join effectively.

Re: proc sql-Join

Ronein — Wed, 25 Jul 2018 09:57:31 GMT

Thanks a lot.]

So it is just another way to write inner join .(using comma instead of writing inner join and using where instead of on).

It is an equivalent program

proc sql;
   title1 'Survey Responses';
   select a.Answer, a.State,
                  count(*) as Count,
                  calculated Count/Subtotal as Percent format=percent8.2
   from survey  as  a 
         inner join (select Answer, 
                 count(*) as Subtotal
                 from survey
                  group by Answer) as b
   on a.Answer=b.Answer
   group by a.Answer, a.State;
quit;

Re: proc sql-Join

RW9 — Wed, 25 Jul 2018 10:05:49 GMT

Yes, should be. You can always try it...