topic Why the split option is not working in proc sgplot? in Graphics Programming

Why the split option is not working in proc sgplot?

Lee_wan — Wed, 04 Sep 2019 02:52:50 GMT

Hi all,

I am trying to split the xasix values in two lines with PROC SGPLOT but it is not working.

Thanks for your idea.

proc sgplot data=trt;
  scatter x=vis y=mean/group=trt01an yerrorlower=lower yerrorupper=upper markerattrs=(symbol=circlefilled) ;
  series x=vis y=mean/group=trt markers markerattrs=(symbol=circlefilled) name='a';
  keylegend 'a'/location=outside position=BOTTOM;
  xaxis values=(1 2 3 4 5 6 7 8) VALUESDISPLAY=("Day 2" "Day 4" "Day 6" "Day 8" "Day 10" "24 hour post*the last dose" 
  "Exit*(Day 14)" "Follow up*(Day 17)")  FITPOLICY=SPLIT SPLITCHAR="*"  label='Planned Time Point' offsetmin=0.05 offsetmax=0.05;
  yaxis label='Mean of Change from Baseline with 95% CI';
run;

Re: Why the split option is not working in proc sgplot?

PeterClemmensen — Wed, 04 Sep 2019 06:30:44 GMT

Hi and welcome to the SAS Communities 🙂

What are you trying to achieve? Can you explain a bit further what your desired result looks like?

Re: Why the split option is not working in proc sgplot?

DanH_sas — Wed, 04 Sep 2019 13:54:19 GMT

I believe the reason that you're getting the behavior you're getting is that a FITPOLICY of SPLIT does not work on a LINEAR axis. Try doing the following:

1. Since you're placing your data in equidistant locations, set TYPE=DISCRETE on the XAXIS.

2. It appears that you always want to split on the "*", so use FITPOLICY=SPLITALWAYS instead of SPLIT. The SPLIT policy will happen only if there is a tick value collision while fitting.

3. On the SCATTER and SERIES, set GROUPDISPLAY=CLUSTER. This will cluster each of your groups around each X value. Make sure your "vis" values are only 1-8, as it appears like the X values might be adjusted slightly to get the look in your picture.

Hope this helps!
Dan

Re: Why the split option is not working in proc sgplot?

Lee_wan — Fri, 06 Sep 2019 05:43:07 GMT

Amazing！ It's working!

Thank you very muck!

You're a Picasso in the SAS figure.