Proc compare

[Ranjeeta]

proc compare base=HT_ORIG_2017_subset  compare=HT_NEW_2017_subset out=check outnoequal outdif outstats=diffstat noprint;

id CIHI_KEY;

var HOMETIME AWAYTIME ;

   with HOMETIME AWAYTIME;

  

   title 'Comparison of Variables in Different Data Sets';

run;

I get  the check dataset with 34 obs – are these 34 obs the only obs that have a different value for the 2 var variables specified in the var statement  in the 2 datasets?

And the diffstat has the stats produces for the 2 var variables for the 678 obs in the 2 datasets  specified in main and compare

How would I instead produce the same stats for the 34 obs ?

[blueskyxyz]

data data1_dif;
	merge check(where=(base) in=a) data1;
	by keyvars;
	if a;
run;

/*pick the different obs*/
data data2_dif;
	merge check(where=(base) in=a) data2;
	by keyvars;
	if a;
run;

/*compare indirectly*/
proc compare b=data1_dif c=data2_dif  OUT=OUTCOMP LISTALL OUTBASE OUTCOMP OUTDIFF OUTNOEQUAL OUTSTATS=diffstat;
run;

[Ranjeeta]

Hello

What is the where base variable?
Are data 1 and data 2 the 2 datasets that I am comparing
Is the check dataset the one that I created using the code above?

[blueskyxyz]

/*1. first compare directly*/
proc compare base=data1  compare=data2  OUT=OUTCOMP1 LISTALL OUTBASE OUTCOMP OUTDIFF OUTNOEQUAL OUTSTATS=diffstat;
id CIHI_KEY;
var HOMETIME AWAYTIME ;
with HOMETIME AWAYTIME; /*can drop this line since variable name is same as var statements*/
run;

/*pick the different obs*/
/*keyvars: CIHI_KEY， to distinguish each unique record in your datasets */

data data1_dif;
	merge OUTCOMP1 (where=(base) in=a) data1;
	by keyvars;
	if a;
run;

/*pick the different obs*/
data data2_dif;
	merge OUTCOMP1 (where=(base) in=a) data2;
	by keyvars;
	if a;
run;

/*2. second compare indirectly*/
proc compare b=data1_dif c=data2_dif  OUT=OUTCOMP2 LISTALL OUTBASE OUTCOMP OUTDIFF OUTNOEQUAL OUTSTATS=diffstat;
id CIHI_KEY;
var HOMETIME AWAYTIME ;
run;

Yes, it is better if you can upload sample datasets to testthe code