topic Overlaying a panels with line plots in Graphics Programming

Overlaying a panels with line plots

Melk — Mon, 04 Dec 2017 21:01:04 GMT

I have a vertical bar plot of means by visit number paneled by group.

Since I only have a few patients in my sample, I would like to overlay this plot with individual line plots. Is there an easy way to do this?

Re: Overlaying a paneled vertical bar chart with line plots

PeterClemmensen — Mon, 04 Dec 2017 07:29:00 GMT

Use the SERIES Statement to plot your individual IDs and use the VBARPARM instead of VBAR, since SERIES and VBAR cannot be overlayed.

Also check out Plot Type Compatibility for a nice summary of what plot types can be overlayed in PROC SGPLOT

Re: Overlaying a paneled vertical bar chart with line plots

PeterClemmensen — Mon, 04 Dec 2017 07:36:30 GMT

Be aware though, that with VBARPARM, you have to calculate the mean values beforehand.

Also, provide some sample of your data to get a code answer 🙂

Re: Overlaying a paneled vertical bar chart with line plots

Melk — Mon, 04 Dec 2017 18:03:41 GMT

Sure and thanks!

Sample data:

ID Visit Group Y

1 1 0 20

2 1 0 24

3 1 1 33

4 1 1 15

5 1 0 26

6 1 1 25

1 2 0 16

2 2 0 12

3 2 1 53

4 2 1 23

5 2 0 10

6 2 1 29

Re: Overlaying a paneled vertical bar chart with line plots

PeterClemmensen — Tue, 05 Dec 2017 09:22:19 GMT

If I understand you correctly, then do something like this

data have;
input ID$ Visit Group Y;
datalines;
1 1 0 20
2 1 0 24
3 1 1 33
4 1 1 15
5 1 0 26
6 1 1 25
1 2 0 16
2 2 0 12
3 2 1 53
4 2 1 23
5 2 0 10
6 2 1 29
;

proc sql;
	create table help as
	select *
	      ,mean(Y) as VisitMean
	from have
	group by Group, Visit
	order by Group, Visit, ID;
quit;

title "Panel Overlayed With Line Plots";
proc sgpanel data=help;
	panelby Group;
	vbarparm category=visit response=VisitMean;
	series x=Visit y=y / group=ID markers markerattrs=(symbol=circlefilled);
	rowaxis display=(nolabel) grid;
	colaxis display=(nolabel);
run;

Re: Overlaying a paneled vertical bar chart with line plots

Melk — Tue, 05 Dec 2017 18:17:58 GMT

worked beautifully, thank you!