Proc Append

[jacksonan123]

I have two data sets which I have produced from the following macro code

PROC MEANS DATA=MERGFIN&K noprint;
VAR KL&K ;
OUTPUT OUT=KLALL&K   sum=skl&K ;
RUN;

The data sets klall1 :

_type_     _freq_     skl1

0               22           0.41

and klaa2

_type_     _freq_     skl1

0               22           0.47

When I use the append code

PROC APPEND BASE=KLALL DATA=KLALL&K  FORCE;
*PROC PRINT;
RUN;

I get the following warning:

NOTE: Appending WORK.KLALL2 to WORK.KLALL.

WARNING: Variable skl2 was not found on BASE file. The variable will not be added to the BASE file.

WARNING: Variable skl1 was not found on DATA file.

NOTE: FORCE is specified, so dropping/truncating will occur.

NOTE: There were 1 observations read from the data set WORK.KLALL2.

NOTE: 1 observations added.

How can I get the following final data set KLALL that I want?

KLALL:

_type_     _freq_     skl1

0               22           0.41

0               22           0.47

[PaigeMiller]

The data sets klall1 :

_type_     _freq_     skl1

0               22           0.41

and klaa2

_type_     _freq_     skl1

0               22           0.47

Your data, as presented, does not match what SAS is telling you. There is no SKL2 in the data you present, yet SAS finds it an decides this is an error. Could you please clarify?

[s_lassen · Accepted solution]

I think you misread the macro code and the variable headers in your KLALL2 table. The summary variable is not called SKL1, but SKL2, that's the problem.

The easy way to solve the problem seems to be to change the SUM= part of the OUTPUT statement in PROC MEANS:

OUTPUT OUT=KLALL&K   sum=skl1 ;

That way, the summary variable will be named SKL1 in all your temporary outputs.

[ballardw]

This may be a silly question but after you have all of these proc means results appended together how do you know which data set any specific value came from? Or the original variable?

Relying on order of appearance in a data set to indicate source or other properties is generally a very poor idea.

[jacksonan123]

A summary confidence interval will be calculated so data order is not important.