topic Re: proc gchart (group % and overall % side by side) in SAS Programming

proc gchart (group % and overall % side by side)

wave43 — Thu, 03 Mar 2016 14:18:14 GMT

I am creating a bar chart where I display 30 individual charts using the "by statement":

proc gchart data = analytic_file;

by group_id;

vbar age_category / type=percent;

run;

The above creates 30 individual charts (for each group) showing the age category percentages for each group_id (great!)

On each of the 30 charts, I want to show side by side 1) the age_category precentages for the group AND 2) the overall percentages for the entire population (analytic_file, not grouped by groupid). Ex. There are 30% who are 18-25 in groupd A, and 45% who are 18-25 for the entire population.

How would I do this?

Re: proc gchart (group % and overall % side by side)

RW9 — Thu, 03 Mar 2016 14:40:13 GMT

Well, advice one - move to sgplot/graph template language. With this you can overlay as much as you want. You can find examples for almost any type of graph here:

http://blogs.sas.com/content/graphicallyspeaking/

If you can't find what you want there then post some test data (datastep), and what the output should look like.

Re: proc gchart (group % and overall % side by side)

SAS_inquisitive — Thu, 03 Mar 2016 14:48:32 GMT

Here is handy cheat sheet for GTL.

https://support.sas.com/rnd/app/ODSGraphics/TipSheet_GTL.pdf

Re: proc gchart (group % and overall % side by side)

wave43 — Thu, 03 Mar 2016 15:42:57 GMT

proc sgplot was defintely the way to go, and solved a couple of my other problems.

However, I still have the same questions. I want to overlay (or side by side) each of these 35 graphs generated with the by statement:

proc sgplot data= analytic_cohort;
by pat_groupid;
vbar agecat_clinical;
run;

With the one graph generated by:

proc sgplot data= analytic_cohort;
vbar agecat_clinical;
run;

B) contains the overall agecat_clinical for the entire population and I want that next to the age_cat clinical for the pat_groupid.

Re: proc gchart (group % and overall % side by side)

wave43 — Thu, 03 Mar 2016 17:03:44 GMT

I figure it out. I am using proc freq to come up with percentages for each of the groups individually:

http://blogs.sas.com/content/iml/2012/12/03/show-percentages-for-bar-charts-with-proc-sgplot.html

Then I am merging the output of the two datasets for two new variables: one for the percent overall, and one for the percent for the group.