Proc Tabulate

[yadalsa]

Hi Team,

Iam using Proc Report for creating a table. I have created a small dataset. i have 4 levels of a variable (L,P,S,W) and then sub_types within each level

O,V,U

the desired output is attached. Iam trying to find frequencies of only the 1's, Iam not interested in the 0's

Also, is there a way to output the data in a desired fashion, for eg.,

i want par6, par7 and part8 to be followed by part1 etc.,

Thank you,

[Reeza]

You should post some sample data in addition. Also, please don't double post your question to multiple forums.

You won't get a dataset in that fashion, but you will get output displayed in that manner. Please clarify which is your desired output, I'm assuming the display since you can't have multiple headers for variables.

[yadalsa]

Hi Reeza,

I created a dataset and an output but i was not able to post 2 spreadsheets. i have the dataset and the output in one spreadsheet in different tabs. Hopefully this should give us a better picture. 

Thank You.

[Astounding]

When you have a variable that is always 0 or 1, change the approach.  To get a single statistic per variable:

To get the frequencies of the "1" values, get the sum.

To get the percentage of "1" values, get the mean.

You may have to adjust your computations, if your variable could take on other values, such as a missing value.

Good luck.

[yadalsa]

Thank You very much.

[Reeza]

1. As mentioned, when you have 0/1 you can sum
2.To control order you need to create a variable that corresponds to the order and then use that to control the order.
3. You didn't answer the question - do you need this as output or a dataset?

[yadalsa]

hi Reeza,

i think i figured it out, i ran a means within the table statement. 

Thank you

proc tabulate data = test out = test1; by procedure1;

var par1 par2....;

class type sub_type;

table all*n (par1 par2 ) *mean = '' *f=pct100fmt.

, type = '' * sub_type= ''  /nocellmerge;

run;

[Reeza]

Yay! Glad to see it worked for you 🙂