proc summary regional hierarchy

[kubbo]

Hi everyone,

I'm having an issue with proc summary in SAS 9.4. My code is:

proc summary data=project.new nway completetypes;
 class product_group product_id region district sales_channel status;
 var count premium age_policyholder_analysis_date_G duration_G;
 output out=project.test1
 sum(premium muzi zeny)=sum_premium muzi zeny
 mean(age_policyholder_analysis_date_G duration_G)=mean_age  mean_duration;
run;

I'm trying to create an "aggregation" but the problem is when using completetypes statement. I get combination of each variables - also combination of "region" and "district" variables which makes no sense because for example distrct "Bratislava" doesn't exist in region "Banskobystrický" (see attached screenshot).

But when I do not use completetypes statement I don't get missing sums (in district "Brezno" for example) which I need.

Is there a way to use some sort of regional hierarchy or something?

Thank you.

[jklaverstijn]

What about eliminating REGION from your summary? You can later join it back in by means of a lookup table or a format that maps regions to districts. Same could apply to the product_group-porduct_id hierarchy.

Hope this helps,

- Jan.

[kubbo]

Yes. It's interesting and simple solution. I'm working on it right now.

I really appreciate your help.

[Reeza]

If you need specific combinations you can use CLASSDATA to specify the levels you're interested in, using PROC MEANS.