topic Re: Crosstables with arrays / several conditions in SAS Programming

Crosstables with arrays / several conditions

dennisl — Mon, 01 Apr 2013 18:34:38 GMT

Dear SAS users,

I want to make a Crosstable with a character variable called fab and a set of variables, for which I made an array. The variables in the array all contain names with the chain icd_nd* where star is a number from 1 to 89.

The crosstable should contain only observations where the condition of another variable, which is a dummy, having the value 1 is fulfilled (gdrg = 1). Another condition should be that that only a set of approximately 30 values, which appear in icd_nd* should be accounted for in the crosstab. For example, I want to look whether the value "test" is in all icd_nd* and crosstab it with fab. Then I want to look if "test1" is in icd_nd* and crosstab it, independent whether "test" is in icd_nd*.

So I have two conditions, gdrg=1 and whether a set of values is in icd_nd*. I could use "where (gdrg =1);" for the first condition, but I dont know if the solution for the latter problem would have an impact on this syntax. Can you help me with the solution for my latter problem?

Best,

Dennis

Re: Crosstables with arrays / several conditions

PGStats — Mon, 01 Apr 2013 18:57:59 GMT

Please give us a small example of input and expected output data. - PG

Re: Crosstables with arrays / several conditions

dennisl — Mon, 01 Apr 2013 20:21:52 GMT

Input:

Dummy variable gdrg

set of alphanumeric variables icd_n* where star is a number from 1 to 100. Variables have values "test","test1",test2",...., "test100"

character variable fab with values "fab1", "fab2",...."fab100"

Output:

A table counting all observations, excluding gdrg =0;

Columns: fab1 to fab100

Rows: test1, test10, test50, test70

The thing is, test1,10,... can occur in all icd_n* variables. If the icd_n* value, which is selected for the rows, occurs together with another one of the selected row values within one observation but in different icd_n* variables, they should be accounted for separately. Hope that this explanation is more clear.

Re: Crosstables with arrays / several conditions

PGStats — Tue, 02 Apr 2013 02:26:48 GMT

Something like this?

data have;
input gdrg (icd_n1-icd_n5) ($) fab $;
datalines;
1 test1 test10 test test70 test50 fab51
1 test15 test10 test test70 test50 fab4
0 test10 test10 test test50 test15 fab32
1 test1 test10 test15 test70 test50 fab4
1 test1 test15 test test70 test50 fab10
;

data haveList(keep=t fab);
set have;
array icd_n{100};
if gdrg;
do i = 1 to dim(icd_n);
     t = icd_n{i};
     if not missing(t) then output;
     end;
run;

proc sql;
create table wantList as
select t, fab, count(*) as n
from haveList
group by t, fab
order by t;
quit;

proc transpose data=wantList out=want(drop=_name_);
by t;
var n;
id fab;
run;