topic Re: Grouping observations using multiple BY Variables in SAS Programming

Grouping observations using multiple BY Variables

souji — Sun, 05 Jan 2020 02:47:45 GMT

Can you please make me understand , how the second by variable read it out put : if first.y ;

X	Y	Z
1	A	27
1	A	33
1	B	45
2	A	52
2	B	69
3	B	70
4	A	82
4	C	91

The following SAS program is submitted:

data two;
   set one;
   by x y;
   if first.y;
run;

proc print data=two noobs;
run;

What is output look like?

Re: Grouping observations using multiple BY Variables

unison — Sun, 05 Jan 2020 04:22:10 GMT

Here's the full code.

data one;
	input x y $ z;
	datalines;
1 A 27
1 A 33
1 B 45
2 A 52
2 B 69
3 B 70
4 A 82
4 C 91
;
run;

proc sort data=one;
	by x y;
run;

data two;
	set one;
	by x y;

	if first.y;
run;

proc print data=two;
run;

You're going to get only the first observation of each combination of x,y. In this example, the 2nd observation will not be output (since x=1, y=A appears in the 1st observation).

To better understand what is happening with first.y, I'd recommend submitting the following and inspecting fy:

data two;
	set one;
	by x y;

	fy = first.y;
/* 	if first.y; */
run;

proc print data=two;
run;

Re: Grouping observations using multiple BY Variables

SASKiwi — Sun, 05 Jan 2020 04:22:51 GMT

@souji - This question is explained in the documentation: https://documentation.sas.com/?docsetId=lrcon&docsetTarget=n01a08zkzy5igbn173zjz82zsi1s.htm&docsetVersion=9.4&locale=en

You will learn more quickly if you check the documentation first, and then if the answer is still not clear, post in the Communities.

Also questions like "What is output look like?" you can answer yourself by running the program.

Re: Grouping observations using multiple BY Variables

Kurt_Bremser — Sun, 05 Jan 2020 08:31:09 GMT

@souji wrote:

What is output look like?

See Maxim 4.

Re: Grouping observations using multiple BY Variables

ed_sas_member — Sun, 05 Jan 2020 09:23:55 GMT

Hi @souji

Hereafter is what First. and Last. variables look like if you group your data by X and then by Y:

-> for X, the first variable of the BY statement is quite intuitive -> for each new value of X, in the order shown, the first.X variable take the value 1. Otherwise, it is set to 0.

-> for the Y variable, for each new group of X and each new value of Y inside those groups, the first.Y variable is set to 1. Otherwise, it is set to 0.