topic Re: Add a reference line PROC SGSCATTER in Graphics Programming

Add a reference line PROC SGSCATTER

marisa_jenkins — Thu, 06 Dec 2018 01:09:27 GMT

HI all

I know you can add a refline in PROC SGPLOT but can you do it in PROC SGSCATTER?

e.g. This is the code we would use to get a refline at Y=0 using PROC SGPLOT but we are teaching a course that primarily uses SGSCATTER as has the ability to produce multiple scatterplots in one PROC and keen to just stick with that.

PROC SGPLOT DATA=dataset;
SCATTER X= xvar Y= yvar;
LOESS X= xvar Y= yvar;
REFLINE 0;
RUN;

Thank you Marisa

Re: Add a reference line PROC SGSCATTER

Reeza — Thu, 06 Dec 2018 03:58:06 GMT

Not AFAIK but you can add a regression or loess line.

Sgcatter does have tmplout on the PROC statement so you can get the GTL code and modify that.

Or switch to SGPANEL?

Re: Add a reference line PROC SGSCATTER

marisa_jenkins — Thu, 06 Dec 2018 05:24:56 GMT

Thank you Reeza

Yes we can add a LOESS line in PROC SGSCATTER but are looking for a ref line at Y = 0.

We currently use a mix of PROC GPLOT and PROC SGSCATTER but are hoping to just use one and not introduce another to be consistent with out other units.

Marisa

Re: Add a reference line PROC SGSCATTER

PGStats — Thu, 06 Dec 2018 06:08:15 GMT

Could also be done with SG annotation.

Re: Add a reference line PROC SGSCATTER

marisa_jenkins — Wed, 12 Dec 2018 00:51:21 GMT

Thank you PG

We are keen to keep with PROC SGSCATTER as that is what is used throughout our course.

Marisa

Re: Add a reference line PROC SGSCATTER

PGStats — Wed, 12 Dec 2018 05:00:17 GMT

SGSCATTER is not a good choice as a general purpose graphing procedure, if I may say so. Here is how to draw reference lines with annotation in SGSCATTER:

data reflines;
retain drawspace "layoutpercent" linecolor "blue";
input function $ x1 y1 x2 y2;
datalines;
line 11 55.5 53.5 55.5 
line 56 55.5 99   55.5
;

proc sgscatter data=sashelp.class sganno=reflines;
compare x=(age height) y=weight / loess=(degree=2);
run;

Positioning the reference lines was by trial and error as you cannot tie their positions to data values.