topic Re: How to do a inner join in SAS Enterprise Miner in SAS Data Science

How to do a inner join in SAS Enterprise Miner

PruthviBhupathiraju — Fri, 01 Aug 2014 16:58:24 GMT

I tried using merge node.But, one data set contained 38,000 records , other data set contained 40,000 records and resultant data set contains 40,000 records. However it should contain 38000 records or less when we do inner join. So can somebody please let me know how to do inner join

Re: How to do a inner join in SAS Enterprise Miner

YujunChiang — Fri, 08 Aug 2014 06:36:17 GMT

How about open the program editor and use the SQL procedure?

Re: How to do a inner join in SAS Enterprise Miner

Kurt_Bremser — Fri, 08 Aug 2014 07:08:23 GMT

Or, in a code node, use (x being your merge variable(s))

proc sort data=A;

by x;

run;

proc sort data=B;

by x;

run;

data C;

merge

A (in=in_a)

B (in=in_b)

;

by x;

if in_a and in_b;

run;

Re: How to do a inner join in SAS Enterprise Miner

jakarman — Fri, 08 Aug 2014 08:18:54 GMT

Eminer is for predictive analytics supporting that all in logical steps.

Data preparation should be done as a separate step before. With the Eminer license you get also Eguide and JMP to be able to that.

Of course you can use the eminer code node but why not use tools better suited for the question?

Re: How to do a inner join in SAS Enterprise Miner

stat_sas — Fri, 08 Aug 2014 10:55:53 GMT

Hi,

Why you are expecting 38000 records? What if you have multiple records in 40000 dataset for joining key.

Re: How to do a inner join in SAS Enterprise Miner

WendyCzika — Thu, 30 Nov 2017 16:34:43 GMT

We came across this one during our "Jam session" today, and just wanted to offer a minor tweak to the solution given by KurtBremser above (thank you KurtBremser!)... in order to have that data set exported to a subsequent node after the SAS Code node, you can do the following:

proc sort data=A;

by x;

run;

proc sort data=B;

by x;

run;

data &em_export_train;

merge

A (in=in_a)

B (in=in_b)

;

by x;

if in_a and in_b;

run;