topic First variable of dataset in SAS Programming

First variable of dataset

Laylitta — Wed, 20 May 2020 09:58:07 GMT

Hello.

I need to sort the variables by mif, ddstl, src_ref. But then I need to output from this dataset only the first variable by mif and src_ref.

Can you help me? I tried this, but not the result which I want

proc sort data=work.QUERY_FOR_ACTIVITY_0008 out=work.skuska;
by mif descending ddstl src_ref;
run;

data work.pokus;
set WORK.skuska;

by mif src_ref;
if first.src_ref then output;

run;

Here is table - I want this yellow rows only:

src_ref can be more then twice in the table, but only with very similar time- difference 40sec.

Re: First variable of dataset

ed_sas_member — Wed, 20 May 2020 09:44:43 GMT

Hi @Laylitta

There is no yellow rows 😉

Could you please update your post with the last part?

Re: First variable of dataset

Kurt_Bremser — Wed, 20 May 2020 09:45:52 GMT

Change the BY statement in the sort to this:

by mif src_ref descending ddstl;

Re: First variable of dataset

Laylitta — Wed, 20 May 2020 09:48:34 GMT

I already updated, I dont know, why it doesn´t upload - I tried it twice, but now I hope, you will see the picture.

Re: First variable of dataset

ed_sas_member — Wed, 20 May 2020 09:52:28 GMT

Hi @Laylitta

We can see the rows now 😊

Please try this:

by mif descending src_ref;
if last.src_ref then output;

Best,

Re: First variable of dataset

Laylitta — Wed, 20 May 2020 09:53:23 GMT

This solution will not works for me. Because the key is ddstl. The observations are 2 together with just the differenct 40sec, but I need just one from them. If I will sort the first by src_ref, they can be e.g. 10 observations. The first two will have the date and time e.g. 5may10:15:30 and 5may 10:16:10 and the other two will have 8may10:30:40 and 8may10:31:20.

Re: First variable of dataset

RichardDeVen — Wed, 20 May 2020 10:53:31 GMT

You will need to use LAG() because you are ignoring a sorted level. There is a possibility your expectations of when src_ref changes will be subverted, especially if src_ref 'bounces around' on a given date.

Example:

data want;
  set have;
  by mif;
  if first.mif or src_ref ne lag(src_ref);
run;

Re: First variable of dataset

Kurt_Bremser — Wed, 20 May 2020 12:01:12 GMT

Why all the hassle with pictures? If you posted your data AS RECOMMENDED(!) in a data step with datalines, it only takes a copy/paste into a "little running man" window. No separate uploading necessary. And you can mark lines in the text with additional "data" that is not read.