SGPLOT VBOX Change Category Color

[acuffza]

I'm using SAS 9.4. I'm creating box plots with two categories and I would like to choose a different, customized color for each category. If possible, I don't want to create a graph template because these graphs are exactly how I need them to be and I only want to change the color. Here's my code:

%macro differences_box_plot (metabolite=, metabolite_title=, format=);

ods graphics / width=1.5in height=1.1in border=off;
proc sgplot data=diffs;
vbox &metabolite. / category=num_drawnum_phen fillattrs=(color=darkgray) whiskerattrs=(color=black) medianattrs=(color=black) meanattrs=(color=black);
title height=12pt "&metabolite_title." ;
xaxis discreteorder=data display=(nolabel) valueattrs=(size=7pt);
yaxis labelattrs=(size=7pt) valueattrs=(size=7pt);
label &metabolite.="Δ [metabolite], mM";
refline 0 / lineattrs=(pattern=shortdash);
format num_drawnum_phen &format..;
run;
title "";

%mend differences_box_plot;

%differences_box_plot (metabolite=Adipate, metabolite_title=Adipate, format=sens_sig);

[Reeza]

Assuming SAS 9.3+ you can try an attribute map instead, it's much easier than modifying the template.

https://blogs.sas.com/content/iml/2012/10/17/specify-the-colors-of-groups-in-sas-statistical-graphic...

[DanH_sas]

Add:

group=num_drawnum_phen

to your VBOX statement and you should get independent colors. As for controlling the colors, do you want specific colors assigned to your data values, or does it matter?

Thanks!
Dan

[acuffza]

There are specific colors I'd like to make each group, it's defaulting to red and blue for the outlines now (other colors are the grays I assigned, but I mainly want to change the fill color of each group). That also made a legend appear which I don't want, do you know how to get rid of that?

[DanH_sas · Accepted solution]

To make the legend go away, add NOAUTOLEGEND to the SGPLOT statement.

To assign the fill colors, use an attributes map (like Reeza mentioned). Here is a simple example:

data attrmap;
retain id "myid" linecolor "gray";
length value $ 1 fillcolor $ 4;
input value $ fillcolor $;
cards;
F pink
M blue
;
run;

proc sgplot data=sashelp.class dattrmap=attrmap noautolegend;
vbox weight / category=sex group=sex attrid=myid;
run;

Hope this helps!

Dan

[DanH_sas]

You might also want to add an entry for MARKERCOLOR in your attributes map if you are displaying outliers.

[acuffza]

That got me a lot closer, but it's still choosing red and blue for some reason! Here's my modified code:

data diff_attrmap;
retain id "myid" linecolor "gray"; length fillcolor $4;
input value fillcolor $;
cards;
1 lime
2 cyan
;
run;

%macro differences_box_plot (metabolite=, metabolite_title=, format=);

ods graphics / width=1.5in height=1.1in border=off;
proc sgplot data=diffs dattrmap=diff_attrmap noautolegend;
vbox &metabolite. / category=num_drawnum_phen group=num_drawnum_phen attrid=myid/*fillattrs=(color=darkgray) whiskerattrs=(color=black) medianattrs=(color=black) meanattrs=(color=black)*/;
title height=12pt "&metabolite_title." ;
xaxis discreteorder=data display=(nolabel) valueattrs=(size=7pt);
yaxis labelattrs=(size=7pt) valueattrs=(size=7pt);
label &metabolite.="Δ [metabolite], mM";
refline 0 / lineattrs=(pattern=shortdash);
format num_drawnum_phen &format..;
run;
title "";

%mend differences_box_plot;

[acuffza]

Nevermind, I fixed it! I saw that "value" had to be character. Thank you very much!

[DanH_sas]

Yes, the VALUE column must contain the FORMATTED value of the group.