combining duplicates

[milts]

hi! how can you combine duplicates(not summing them up) using data step or proc sql? advance thanks! 🙂

[sbb]

What is meant by "combine"? A SAS MERGE (using a BY) within a DATA step or a PROC SQL with a JOIN operation in a SELECT, come to mind. You mentioned specifically that no "summing" was involved.

Maybe explain in more detail the objective with combining duplicates -- also, consider that if there are duplicates (multiple as in more than 2 obs/rows with the same BY key variables, how will this condition be handled with combining rows?

Need more detail about your objective -- show some BEFORE and AFTER data rows in a post/reply.

Scott Barry
SBBWorks, Inc.

[milts]

anyway i got it now. just use the nodup on proc sort 😃

[sbb]

Keep in mind the NODUPKEY (limited to BY variable list) option in addition (requires duplicate observations to be adjacent for them to be removed, based on the BY list granularity).

And, if you want to preserve the duplicate observations, there is a DUPOUT= keyword.

Scott Barry
SBBWorks, Inc.

[sbb]

Please disregard the incomplete prior reply...let's try this again:

Keep in mind the NODUPKEY (limited to BY variable list) option. Also, note that the NODUP option requires duplicate observations to be adjacent for them to be removed, which would be based on the BY list granularity.

And, if you want to preserve the duplicate observations, there is a DUPOUT= keyword.

Scott Barry
SBBWorks, Inc.

[milts]

got that 😃 thanks for the additional information on nodupkey and dupout.