Getting total of variable to use in new variable calculation

[JethrowCatch]

Hello,

I am using a dataset with 12 numeric variables. I need to use a total of one variable, x12, to calculate every observation's percentage of the whole, and I would like to do this within the DATA step.

DATA example;
INPUT x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12;
x13 = (x12 / total of x12) * 100;

DATALINES;
... ... ... ... ... ... .. .. .. 200
... ... .. .. .. .. .. ..  .. .. 300

;
PROC PRINT DATA = example;
                        SUM(x12)
                        SUM(x13)
RUN;

Now, I need to obtain the sum of all x12 values and use it in the calculation of a new variable, x13 which is a relative percentage. The formula to calculate x13 is as follows:

x13 = (current x12 value / total sum of all x12 values) * 100

Can anyone help me with obtaining the total sum of all x12 values in the DATA step?

[PaigeMiller]

DATA example;
INPUT x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12;
array x x1-x12;
array pct pct1-pct12;
rowsum=sum(of x1-x12);
do i=1 to 12;
    pct(i) = (x(i) / rowsum) * 100;
end;
datalines;

;
run;

[JethrowCatch]

Hello @PaigeMiller ,

Thank you for your swift response. I think you misunderstood my question. I need to obtain the sum of every value in the column x12, not each row.

[Reeza]

SQL is faster.

proc sql;
create table want as
select *, sum(x12) as total,  x12 / calculated total as x13
from example;
quit;

Otherwise, you need to:

1. Create a total variable
2. Merge it in to your data set
3. Do the calculation. 

If you really want a data step, see this example on how to merge in averages (SUM) would be the total and you can modify it as needed.

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

---

@JethrowCatch wrote:

Hello,

I am using a dataset with 12 numeric variables. I need to use a total of one variable, x12, to calculate every observation's percentage of the whole, and I would like to do this within the DATA step.

DATA example;
INPUT x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12;
x13 = (x12 / total of x12) * 100;

DATALINES;
... ... ... ... ... ... .. .. .. 200
... ... .. .. .. .. .. ..  .. .. 300

;
PROC PRINT DATA = example;
                        SUM(x12)
                        SUM(x13)
RUN;

Now, I need to obtain the sum of all x12 values and use it in the calculation of a new variable, x13 which is a relative percentage. The formula to calculate x13 is as follows:

x13 = (current x12 value / total sum of all x12 values) * 100

Can anyone help me with obtaining the total sum of all x12 values in the DATA step?

---

[JethrowCatch]

Hello @Reeza ,

Thank you for your response. We have not learned about PROC SQL in school yet. I do not want to seem as if I am cheating by using functions I haven't learned yet.

I have given your 3 step method a go, but I am having some trouble understanding how to obtain the total value before calculating the percentage.

DATA example;
INPUT x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12;
RETAIN TOTAL 0;
TOTAL = sum(TOTAL, x12);

DATALINES;
.....
;

DATA PCT;  			
MERGE  example; BY x1;
x13 =  (x12/TOTAL) * 100;
RUN;	

PROC PRINT DATA = PCT;
RUN:

My problem is that the value of TOTAL increases for each observation. I need for TOTAL to be a variable that has one value, which is the value it takes on after adding the x12 value of the last observation to itself.

[JethrowCatch]

In other words, what I would like to do is to use a specific index of TOTAL in every x13 calculation. However, indexing with curly or straight brackets does not work.

Essentially, the x13 calculation should look as follows:

x13 = ( x12 / TOTAL[30]) * 100;

30 is my number of observations in the example dataset.

Would anyone happen to know how to index observations in SAS?

[Reeza]

Did you work through the example at the link?
Note the PROC MEANS step that calculated the average, you would modify that to calculate the sum.

[JethrowCatch]

Yes, I have. I now obtain the total that I need in the variable TOTALCT. However, the output that I get under the OUTPUT DATA tab is a table showing type, frequency, and total.

The output that I need is a table showing all the original variables x1-x12, with x13 added on. Therefore, I need to use PROC PRINT, but at this moment, nothing shows up under the results tab.

PROC MEANS DATA = EXAMPLE NOPRINT;
    OUTPUT OUT = example SUM(x12) = TOTALCT;
run;
	  
DATA PCT;
MERGE example; BY x1;
x13 = (x12 / TOTALCT) * 100;
RUN;

PROC PRINT DATA = PCT;
RUN;

[Reeza]

Your merge is not correct. Please review the example to see my use of the double SET statement instead.

FYI - The NOPRINT option is specified on the PROC MEANS which suppresses any results. If you'd like to see them, remove the NOPRINT option.