I want to do Roll up and Roll down by my interest

[carl_miles]

Consider the dataset

Data Exam;

format exam_date date9.;
informat exam_date date9.;

Input ID $ exam_date Test1 $;
Datalines;
9000001 01JAN2023    A,B
9000001 15FEB2023      D,B
9000001 20MAR2023   E,C

9000003 15FEB2023      D,B
9000003 20MAR2023   F,C

9000004 20MAR2023   B

9000004 20MAR2023   G

answer1 condition if patient containing A, B, C then only those should in final dataset

my answer1 should be

9000001 A+B+C

9000003 B+C

9000004   B

answer2 condition if patient containing other then A, B, C then only those should in final dataset

my answer2 should be

9000001 D+E

9000003 D+F

9000004 G

please write a code that will help to achieve this result

[Kurt_Bremser]

Looks like some kind of homework or exam. What have you tried so far?

[andreas_lds]

There are many ways to create the datasets you need. But i don't think that posting a solution will help you to solve tasks in the future. So, what have you tried?

[Amir]

Hi,

As others have said, sharing what you have tried will help both you and us. I would also advise sharing the log if you have run the code, including any notes, warnings, or error messages.

Please use the Insert Code icon "</>" when posting code or log content.

Thanks & kind regards,

Amir.

[Kurt_Bremser]

How I would approach it:

• in a DATA step, create two datasets (ans1 and ans2), both containing id and new variable test
• loop through the test1 variable (COUNTW and SCAN functions), and write to ans1 if test is in A,B,C or to ans2 if not
• sort both dataset by id and test with NODUPKEY to remove duplicates
• in two final DATA steps, concatenate the tests (use BY, FIRST. and LAST., RETAIN, CATX)