Assuming you called the macro a "function" out of habit and not because you think will behave the same as a SAS function like LOG() or MEAN() you could just generate one macro call per observation.   So assuming that your dataset is named HAVE you could run something like this: filename code temp; data _null_; set have; file code; put '%fn(input1=' X ',input2=' Y ')' ; run; %include code / source2; The data step will write one line of code per observation.  The %INCLUDE statement will then run the generated lines of code. ... View more

Please include the NObs table. It is possible that you have no valid observations.   The problem could be that you read the data incorrectly. The data you attached seem to be invalid. They look like this: 31-Dec-03 77546 1 0 1 1 31-Dec-03 10104 1 0 1 1 31-Dec-03 77284 1 0 1 1 In your CSV file, there is only one column, which contains values like "31-Dec-03 77546 1 0 1 1" You need to make sure that those numbers are properly read into the variables in your program. Otherwise, the y, x1, and x3 variables might contain missing values.    The other thing you can do is run the following and upload the results: proc print data=test(obs=5); var id date y x1 x3; run; ... View more

I think this example using PROC EXPAND is where you should start. Once you handle the irregular dates, then you can try the rolling regressions.   https://documentation.sas.com/doc/en/pgmmvacdc/9.4/etsug/etsug_expand_examples03.htm ... View more

I'm not following the need for a macro. If you do the merge, then you can do all the regressions with a BY statement. ... View more

thanks i got it ... View more

the sample data here ... View more