Calculate totals and sub-totals

[dstuder]

Given the following sample data

DATA test;
INPUT level :6. name :$100. count :6.;
INFILE DATALINES DSD;
DATALINES;
1, Top-category 1, .
2, Sub-category 1.1, 10
2, Sub-category 1.2, 20
2, Sub-category 1.3, .
3, Sub-sub category 1.3.1, 5
3, Sub-sub category 1.3.2, 5
1, Top-category 2, .
2, Sub-category 2.1, 10
2, Sub-category 2.2, .
2, Sub-category 2.3, 5
;
RUN;

is there a way to calculate the sums of all sub-categories iteratively?

Examples:

• Category 1 should be 30 (10+20+5+5)
• Sub-category 1.3 should be 10 (5+5)

The count-value is only given for the "lowest" categories. If the the value of the "lowest" category is missing "." this should actually be 0 (e.g. sub-category 2.2). This is why top-category 2 should be 15.

[Kurt_Bremser]

Why should top-category 2 have a sum of 10, when 2.1 = 10 and 2.3 = 5? Why would you discard the value of 2.3?

[Shmuel]

1) on your last input line there is a dot instead of a comma after the level.

2) Use next code to create separate variables to define categories and sub categories:

DATA test;
INFILE DATALINES dlm=',' truncover;
INPUT level :6. name :$30. count :6.;
length cat $10;
catx = scan(name,-1,' ');
array ct cat1-cat5;
do i=1 to countw(catx,'.');
  ct(i) = scan(catx,i,'.');
end;
DATALINES;
1, Top-category 1, .
2, Sub-category 1.1, 10
2, Sub-category 1.2, 20
2, Sub-category 1.3, .
3, Sub-sub category 1.3.1, 5
3, Sub-sub category 1.3.2, 5
1, Top-category 2, .
2, Sub-category 2.1, 10
2, Sub-category 2.2, .
2, Sub-category 2.3, 5
;
RUN;

next step run proc summary to calculate totals and subtotals:

proc summary data=test;
 class cat1-cat3;
 var count;
 output out=summary sum=;
run;

[dstuder]

@Shmuel 

Thank you for your solution, which works fine! Unfortunately, I was just giving these numbers in the variable `name` to make my question more understandable here. In reality, my data are like this:

DATA test;
INPUT level :6. name :$100. count :6.;
INFILE DATALINES DSD;
DATALINES;
01, Top, .
0101, Sub, 10
0102, Sub, 20
0103, Sub, .
01031, Sub-sub, 5
01032, Sub-sub, 5
02, Top, .
0201, Sub, 10
0202, Sub, 20
0203, Sub, .
02031, Sub-sub, 10
02032, Sub-sub, .
020321, Sub-sub-sub, 6,
020322, Sub-sub-sub, 4
;
RUN;

So, 0101 should be used instead of 1.1, 02031 instead of 2.3.1,  020311 instead of 2.3.1.1 etc. Do you think you could adapt this for me?

[Shmuel]

So your categories are defined in the LEVEL variable, then

change next lines to separate the parent level from the sub-levels:

instead:

do i=1 to countw(catx,'.');
  ct(i) = scan(catx,i,'.');
end;

use:

1) read level as $6 informat instead a numeric informat

then

array ct cat1-cat3;
do i=1 to length(level)-2 by 2;
   ct(i) = substr(level,i,2);
end;