To create frequency table

[ari]

Dear Community,

I'm trying to create a frequency table using the following sample dataset.

For subjects, var1-var3 are variables measured on different days (day variable). I would like to generate a frequency count for how many subjects with value=1 on any given day per study (startified by study).

Any help to achieve the output is greatly appreciated.

data test;
    input id var1 var2 var3 day stud;
	datalines;
100101	1	1	1	-4	1
100101	1	0	0	1	1
100101	1	0	0	3	1
100101	1	1	1	8	1
100101	1	0	0	15	1
100101	1	0	0	17	1
100101	0	1	1	27	1
100103	1	1	1	-5	1
100103	1	0	0	1	1
100103	1	0	0	3	1
100103	0	1	1	7	1
100106	1	1	1	-4	1
100106	1	0	0	1	1
100106	1	0	0	3	1
100106	1	1	1	8	1
100106	1	0	0	15	1
100106	1	0	0	17	1
100106	1	1	1	26	1
1000101	1	1	1	-13	2
1000101	1	0	0	-8	2
1000101	1	0	0	0	2
1000101	1	0	0	1	2
1000101	1	0	0	1	2
1000101	0	1	1	7	2
1000101	1	0	0	8	2
1000101	1	1	1	15	2
1000101	1	0	0	111	2
1000103	1	1	1	-18	2
1000103	1	0	0	-9	2
1000103	1	0	0	0	2
1000103	0	1	1	6	2
1000103	1	0	0	7	2
1000103	1	0	0	17	2
1000103	0	1	1	19	2
1000103	1	0	0	24	2
4900201	1	0	0	-9	2
4900201	0	1	1	-9	2
4900201	1	0	0	-8	2
4900201	1	0	0	0	2
4900201	1	0	0	1	2
4900201	1	0	0	2	2
4900201	0	1	1	6	2
4900201	1	0	0	7	2
4900201	1	0	0	15	2
;
run;

Any help to achieve the example output is greatly appreciated.

	day	day	day	day	day	day	day
	-4	1	3	8	15	17	27
var1
var2
var3

[Reeza]

PROC MEANS, Example 2 & 10 in the documentation- and since you're using 1 and 0 you can use SUM, MEAN and N statistics for what you want.

SUM = Number of Value = 1

Mean = Percent = 1

N = Number of responses, 0 & 1's.

http://documentation.sas.com/?docsetId=proc&docsetTarget=n1qnc9bddfvhzqn105kqitnf29cp.htm&docsetVers...

[ballardw · Accepted solution]

Proc tabulate allows nesting and grouping in different manners.

See:

data test;
    input id var1 var2 var3 day stud;
	datalines;
100101	1	1	1	-4	1
100101	1	0	0	1	1
100101	1	0	0	3	1
100101	1	1	1	8	1
100101	1	0	0	15	1
100101	1	0	0	17	1
100101	0	1	1	27	1
100103	1	1	1	-5	1
100103	1	0	0	1	1
100103	1	0	0	3	1
100103	0	1	1	7	1
100106	1	1	1	-4	1
100106	1	0	0	1	1
100106	1	0	0	3	1
100106	1	1	1	8	1
100106	1	0	0	15	1
100106	1	0	0	17	1
100106	1	1	1	26	1
1000101	1	1	1	-13	2
1000101	1	0	0	-8	2
1000101	1	0	0	0	2
1000101	1	0	0	1	2
1000101	1	0	0	1	2
1000101	0	1	1	7	2
1000101	1	0	0	8	2
1000101	1	1	1	15	2
1000101	1	0	0	111	2
1000103	1	1	1	-18	2
1000103	1	0	0	-9	2
1000103	1	0	0	0	2
1000103	0	1	1	6	2
1000103	1	0	0	7	2
1000103	1	0	0	17	2
1000103	0	1	1	19	2
1000103	1	0	0	24	2
4900201	1	0	0	-9	2
4900201	0	1	1	-9	2
4900201	1	0	0	-8	2
4900201	1	0	0	0	2
4900201	1	0	0	1	2
4900201	1	0	0	2	2
4900201	0	1	1	6	2
4900201	1	0	0	7	2
4900201	1	0	0	15	2
;
run;

proc tabulate data=test;
   class stud id day;
   var var1 var2 var3;
   /* this does separate table for each level of study*/
   table stud='Study:',
         id,
         day*(var1 var2 var3)*(sum=""*f=best5.)
         /misstext=' '
   ;
   /* nests id within levels of stud*/
   table stud='Study' *  id,
         day*(var1 var2 var3)*(sum=""*f=best5.)
         /misstext=' '
   ;
run;

As @Reeza sum gets the number of 1 values, there are a number of statistics that could be requested as well. Place them inside the () around sum to group as in these tables.

[ari]

.

[Reeza]

Rather than *day try , day in the PROC TABULATE - look at the example again and note where the comma's are placed.