topic Re: Vertical summation by two variables in SAS Programming

Vertical summation by two variables

Ethid — Thu, 12 Jul 2018 17:32:18 GMT

Hello,

I have:

x y z

10 1 67

10 1 430

10 2 122

10 2 51

10 2 219

I want

x y z z_sum

10 1 67 .

10 1 430 497

10 2 122 .

10 2 51 .

10 2 219 392

I am having trouble summing it by two variables. Any ideas? Thanks for any help!

Re: Vertical summation by two variables

Reeza — Thu, 12 Jul 2018 17:39:55 GMT

@Ethid wrote:

I am having trouble summing it by two variables. Any ideas? Thanks for any help!

What troubles are you having? What approach are you taking?

Is that what you want your final output to be, or are you looking for just the summary statistics.

How to summarize data:

https://github.com/statgeek/SAS-Tutorials/blob/master/proc_means_basic.sas

How to add a statistic to your main data set:

https://github.com/statgeek/SAS-Tutorials/blob/master/add_average_value_to_dataset.sas

@Ethid wrote:

Hello,

I have:

x y z

10 1 67

10 1 430

10 2 122

10 2 51

10 2 219

I want

x y z z_sum

10 1 67 .

10 1 430 497

10 2 122 .

10 2 51 .

10 2 219 392

I am having trouble summing it by two variables. Any ideas? Thanks for any help!

Re: Vertical summation by two variables

novinosrin — Thu, 12 Jul 2018 17:41:43 GMT

data have;
input cx      y      z;
cards;
10   1      67
10   1     430
10   2     122
10   2     51
10   2     219
;
data want;
set have;
by cx      y;
retain t;
if first.y then t=z;
else t+z;
if last.y then z_sum=t;
drop t;
run;

Re: Vertical summation by two variables

Ethid — Thu, 12 Jul 2018 17:45:16 GMT

Thank you Novinosrin, this should work!

Re: Vertical summation by two variables

KDG — Thu, 12 Jul 2018 19:13:57 GMT

Note that for your real data, you'll have to add in a proc sort and sort by (in this case) cx and y:

proc sort data=have; by cx y; run;

In this case you didn't get an error because your data was already sorted, but your real data probably isn't and you'll get an error about by variables not being properly sorted.

Re: Vertical summation by two variables

PaigeMiller — Thu, 12 Jul 2018 20:01:56 GMT

I just don't see the point of writing your own data step code to do summation, when SAS already has a PROC that does this, and handles missings and does error checking, all at no extra charge.

proc summary nway data=have;
    class cx y;
    var z;
    output out=want sum=sum_z;
run;