different method

[Radwan]

hello 

i need to make sure about kind of code whether it is correct or not

first i usually use creating a variable (var_lag) to calculate the delta of that var for instance total assets( TA_lag) or inventory 

then i used the statement as follows 

 data real_em;
 set erm_var;
 CFLO_real= (CSFO/TA_lag);
TA1=(1/TA_lag);
S1= (sales/TA_lag);
S2= (delta_sales/TA_lag);
S3= (delta_sales_lag/TA_lag);
S4= (sales_lag/TA_lag);
PRODa= (PROD/TA_lag);
EXPENa= (EXPEN/TA_lag);
run; 

 then i tried to change my way by using this statement directly without creating var_lag

as follows

 data attempt; set erm_var;
 cshflo=csfo/lag(Total_assets);
 TA1=(1/lag(Total_assets));
S1= (sales/lag(Total_assets));
S2= (delta_sales/lag(Total_assets));
S3= (delta_sales_lag/lag(Total_assets));
S4= (sales_lag/lag(Total_assets));
PRODa= (PROD/lag(Total_assets));
EXPENa= (EXPEN/lag(Total_assets));
 run;

of course by first way there were a lot of missing value while in the second way less than, so please which way is correct 

[Kurt_Bremser · Accepted solution]

Using one variable instead of multiple lag() calls renders the same result:

data test1;
set sashelp.class;
s1 = lag(age);
s2 = 1/lag(age);
s3 = weight/lag(age);
run;

data test2;
set sashelp.class;
agelag = lag(age);
s1 = agelag;
s2 = 1/agelag;
s3 = weight/agelag;
drop agelag;
run;

proc compare
  base=test1 
  compare=test2
  criterion=0.00001
  method=RELATIVE
  out=comp(label="Compare Data for TEST1 and TEST2")
  outstats=stat(label="Compare Data Summary Statistics for TEST1 and TEST2")
  outnoequal
  maxprint=50
;
var Name Sex Age Height Weight s1 s2 s3;
with Name Sex Age Height Weight s1 s2 s3;
run;

Dataset comp will be empty, meaning both datasets are identical.