How do I specify the LEGEND statement for these three lines (see code). to look like this but with CLVEF and CPGPRESV and in the box. I made the legend in the picture with mergedlegend "BNPPRONT" "BNPPRONT1";
seriesplot x=xval y=BNPPRONT / name="BNPPRONT" lineattrs=(color=red pattern=dash) yaxis=y /*curvelabel='NTproBNP' curvelabelattrs=(color=red size=6pt)*/;
scatterplot x=xval y=BNPPRONT / name="BNPPRONT1" markerattrs=(color=red symbol=circle) yaxis=y datalabel=BNPPRONT datalabelattrs=(size=6pt);
seriesplot x=xval y=CLVEF / name="CLVEF" lineattrs=(color=blue pattern=dash) yaxis=y2 /*curvelabel='LVEF' curvelabelattrs=(color=blue size=6pt)*/;
scatterplot x=xval y=CLVEF / name="CLVEF1" markerattrs=(color=blue symbol=diamondfilled) yaxis=y2 datalabel=CLVEF datalabelattrs=(size=6pt);
seriesplot x=xval y=CPGPRESV / name="CPGPRESV" lineattrs=(color=blue pattern=dash) yaxis=y2 /*curvelabel='LVOTG' curvelabelattrs=(color=blue size=6pt)*/;
scatterplot x=xval y=CPGPRESV / name="CPGPRESV1" markerattrs=(color=blue symbol=diamond) yaxis=y2 datalabel=CPGPRESV datalabelattrs=(size=6pt);
Here is a simple example of what I was describing:
proc template;
define statgraph merged;
begingraph;
layout overlay;
seriesplot x=ageatstart y=systolic / name="a1" lineattrs=GraphData1;
scatterplot x=ageatstart y=systolic / name="a2" markerattrs=GraphData1;
seriesplot x=ageatstart y=diastolic / name="b1" lineattrs=GraphData2;
scatterplot x=ageatstart y=diastolic / name="b2" markerattrs=GraphData2;
seriesplot x=ageatstart y=weight / name="c1" lineattrs=GraphData3;
scatterplot x=ageatstart y=weight / name="c2" markerattrs=GraphData3;
layout gridded / columns=3 border=true location=outside;
mergedlegend "a1" "a2" / border=false;
mergedlegend "b1" "b2" / border=false;
mergedlegend "c1" "c2" / border=false;
endlayout;
endlayout;
endgraph;
end;
run;
proc sgrender data=sashelp.heart(obs = 5) template=merged; run;
Try this:
Let me know if this works for you. Thanks!
Hi Dan,
Thanks, code below. These statements are producing this legend.
I want total 4 entries, and I thought ADDITIONALNAMES would "merge" the pairs.
%macro m_plate2(pk=0,ticklist=,cmvar=CMDECOD);
%let ticklist=%unquote(&ticklist);
proc template;
define statgraph cardiac_overlay;
begingraph;
layout lattice / rows=1 rowweights=(1 ); /* 70% for cardiac, 30% for CONMED */
layout overlay /
yaxisopts = (
%if &pk %then %do;
label="PK (ng/mL) & NT-proBNP (pmol/L)"
%end;
%else %do;
label="NT-proBNP (pmol/L)"
%end;
labelattrs=(color=red size=9pt) tickvalueattrs=(color=red size=10pt) offsetmin=.01
)
y2axisopts=(label="LVEF (%) & LVOT-G Valsalva (mmHg), (Core Lab)"
offsetmin=.01
labelattrs=(color=blue size=9pt) tickvalueattrs=(color=blue size=10pt) /*linearopts=(viewmin=40 viewmax=70)*/)
xaxisopts=(display=(ticks tickvalues label) label="Week" LABELPOSITION=RIGHT offsetmin=.01 linearopts=(tickvaluelist=(&ticklist) tickvaluefitpolicy=stagger) tickvalueattrs=(size=6pt))
;
%if &pk %then %do;
seriesplot x=xval y=PK / name='PK' lineattrs=(color=red pattern=solid) yaxis=y /*curvelabel='PK' curvelabelattrs=(color=red size=6pt)*/ legendlabel='PK (ng/mL)';
scatterplot x=xval y=PK / name='PK1' markerattrs=(color=red symbol=circlefilled) yaxis=y datalabel=PK datalabelattrs=(size=6pt) legendlabel='PK (ng/mL)';
%end;
seriesplot x=xval y=BNPPRONT / name="BNPPRONT" lineattrs=(color=red pattern=dash) yaxis=y /*curvelabel='NTproBNP' curvelabelattrs=(color=red size=6pt)*/ legendlabel='NT-proBNP (pmol/L)';
scatterplot x=xval y=BNPPRONT / name="BNPPRONT1" markerattrs=(color=red symbol=circle) yaxis=y datalabel=BNPPRONT datalabelattrs=(size=6pt) legendlabel='NT-proBNP (pmol/L)';
seriesplot x=xval y=CLVEF / name="CLVEF" lineattrs=(color=blue pattern=dash) yaxis=y2 /*curvelabel='LVEF' curvelabelattrs=(color=blue size=6pt)*/ legendlabel='LVEF(%) (Core)';
scatterplot x=xval y=CLVEF / name="CLVEF1" markerattrs=(color=blue symbol=diamondfilled) yaxis=y2 datalabel=CLVEF datalabelattrs=(size=6pt) legendlabel='LVEF(%) (Core)';
seriesplot x=xval y=CPGPRESV / name="CPGPRESV" lineattrs=(color=blue pattern=dash) yaxis=y2 /*curvelabel='LVOTG' curvelabelattrs=(color=blue size=6pt)*/ legendlabel='LVOT-G Valsalva (mmHg) (Core)' ;
scatterplot x=xval y=CPGPRESV / name="CPGPRESV1" markerattrs=(color=blue symbol=diamond) yaxis=y2 datalabel=CPGPRESV datalabelattrs=(size=6pt) legendlabel='LVOT-G Valsalva (mmHg) (Core)' ;
innermargin / align=bottom opaque=true backgroundcolor=cxf5f5f5;
axistable x=dosdy value=dose / ;
axistable x=ady value=aval / ;
endinnermargin;
%if &pk %then %do;
mergedlegend "PK" 'PK1' / ADDITIONALNAMES=("BNPPRONT" "BNPPRONT1" "CLVEF" "CLVEF1" 'CPGPRESV' 'CPGPRESV1') title=' ';
%end;
%else %do;
mergedlegend "BNPPRONT" "BNPPRONT1" / ADDITIONALNAMES=("CLVEF" "CLVEF1" 'CPGPRESV''CPGPRESV1') title=' ';
%end;
endlayout;
%if 0 %then %do;
/* ---- CONMED Data Panel ---- */
layout overlay / xaxisopts=(display=NONE offsetmin=.01 linearopts=(tickvaluelist=(&ticklist) tickvaluefitpolicy=stagger))
yaxisopts=(display=(ticks tickvalues)
/*DISCRETEOPTS=(TICKVALUEFITPOLICY=SPLITALWAYS) */
DISCRETEOPTS=(TICKVALUEFITPOLICY=SPLIT TICKVALUESPLITCHAR=';' sortorder=data )
type=discrete reverse=true tickvalueattrs=(color=black size=5pt))
;
/* HiLo Plot for CONMED */
highlowplot Y=&cmvar /*CMDECOD*/ low=s high=e / name="Conmeds" type=line
lineattrs=(pattern=solid thickness=2) /*markerattrs=(symbol=circlefilled)*/;
/* Handling Ongoing Medications */
scatterplot x=o y=&cmvar / markerattrs=(symbol=TriangleRightFilled color=red size=6);
scatterplot x=p y=&cmvar / markerattrs=(symbol=TriangleLeftFilled color=red size=6);
endlayout;
%end;
endlayout;
endgraph;
end;
run;
%mend m_plate2;
No, ADDTIONALNAMES just adds more "single" items to the merged items. The technique I described (multiple merged legends in the layout gridded) should work for you. Give it a try and let me know.
Here is a simple example of what I was describing:
proc template;
define statgraph merged;
begingraph;
layout overlay;
seriesplot x=ageatstart y=systolic / name="a1" lineattrs=GraphData1;
scatterplot x=ageatstart y=systolic / name="a2" markerattrs=GraphData1;
seriesplot x=ageatstart y=diastolic / name="b1" lineattrs=GraphData2;
scatterplot x=ageatstart y=diastolic / name="b2" markerattrs=GraphData2;
seriesplot x=ageatstart y=weight / name="c1" lineattrs=GraphData3;
scatterplot x=ageatstart y=weight / name="c2" markerattrs=GraphData3;
layout gridded / columns=3 border=true location=outside;
mergedlegend "a1" "a2" / border=false;
mergedlegend "b1" "b2" / border=false;
mergedlegend "c1" "c2" / border=false;
endlayout;
endlayout;
endgraph;
end;
run;
proc sgrender data=sashelp.heart(obs = 5) template=merged; run;
Thanks, this example is very helpful.
Looking are your original code gain, I do not think you need to have separate SERIESPLOT and SCATTERPLOT statements. You can display the markers on the SERIESPLOT statement and control the marker attributes. Then, you can use one DISCRETELEGEND for all of your SERIESPLOT statements. For example,
seriesplot x=xval y=PK / name='PK' lineattrs=(color=red pattern=solid) yaxis=y legendlabel='PK (ng/mL)' display=(markers) markerattrs=(color=red symbol=circlefilled) yaxis=y datalabel=PK datalabelattrs=(size=6pt);
John.King ,
Another way is using LEGENDITEM to customize it .
/* Step 1: Create Dummy Dataset */ data dummy_data; input Subjid $ Age_Group $ Treatment $ AVAL; /* Numeric representation for Treatment */ if Treatment = 'DrugA' then trtn = 1; else if Treatment = 'DrugB' then trtn = 2; else if Treatment = 'Refer' then trtn = 3; /* Numeric representation for Age_Group */ if Age_Group = '5-13' then agegrpn = 1; else if Age_Group = '14-21' then agegrpn = 2; else if Age_Group = '22-35' then agegrpn = 3; datalines; S01 5-13 DrugA 6000 S02 5-13 DrugA 7000 S03 5-13 DrugB 8000 S04 5-13 DrugB 9000 S05 5-13 Refer 500 S06 5-13 Refer 12000 S07 5-13 DrugA 10000 S08 5-13 DrugB 11000 S09 5-13 Refer 400 S10 14-21 DrugA 4000 S11 14-21 DrugA 5000 S12 14-21 DrugB 6000 S13 5-13 DrugB 7000 S14 14-21 Refer 1000 S15 14-21 Refer 300 S16 22-35 DrugA 8000 S17 14-21 DrugB 9000 S18 14-21 Refer 100 S19 22-35 DrugA 200 S20 22-35 DrugA 300 S21 22-35 DrugB 400 S22 22-35 DrugB 500 S23 22-35 Refer 50 S24 14-21 Refer 100 S25 22-35 DrugA 600 S26 22-35 DrugB 700 S27 22-35 Refer 150 S28 5-13 DrugA 8000 ; run; proc sort data=dummy_data; by trtn agegrpn; run; proc means data=dummy_data nway noprint; /* by trtn strat2rn;*/ /* var pchg;*/ by trtn agegrpn; var aval; *output out=statall min= max= median= mean= n= q1 = q3 =/autoname; output out=stat n=n mean=mean median=median; run; data final; merge dummy_data stat(drop=_:); by trtn agegrpn; num =1; if TREATMENT='DrugA' then do;mean1=mean;median1=median;end; if TREATMENT='DrugB' then do;mean2=mean;median2=median;end; if TREATMENT='Refer' then do;mean3=mean;median3=median;end; run; proc template; define statgraph boxplot; begingraph / datacolors=(lightblue lightgreen lightgrey) datacontrastcolors=(lightblue lightgreen lightgrey) datasymbols=(trianglefilled squarefilled circlefilled); /* Define Custom Attributes for Age Groups */ discreteattrmap name="groupline" / ignorecase=true; value 'DrugA' / lineattrs=(color=lightblue pattern=1 thickness=2px) markerattrs=(color=lightblue symbol=trianglefilled size=8pt) textattrs=(color=lightblue); value 'DrugB' / lineattrs=(color=lightgreen pattern=1 thickness=2px) markerattrs=(color=green symbol=squarefilled size=8pt) textattrs=(color=lightgreen); value 'Refer' / lineattrs=(color=grey pattern=1 thickness=2px) markerattrs=(color=grey symbol=circlefilled size=8pt) textattrs=(color=grey); enddiscreteattrmap; discreteattrvar attrvar=trtgrp var=trtn attrmap='groupline'; /* Add Custom Legends */ legenditem type=markerline name='A' / lineattrs=(color=blue pattern=1 thickness=2px) markerattrs=(color=blue size=8pt symbol=trianglefilled) label="DrugA" labelattrs=(size=9pt); legenditem type=markerline name='B' / lineattrs=(color=green pattern=1 thickness=2px) markerattrs=(color=green size=8pt symbol=squarefilled) label="DrugB" labelattrs=(size=9pt); legenditem type=markerline name='R' / lineattrs=(color=grey pattern=1 thickness=2px) markerattrs=(color=grey size=8pt symbol=circlefilled) label="Refer" labelattrs=(size=9pt); layout overlay / xaxisopts=(offsetmin=0.12 offsetmax=0.12 label="Age Groups" labelattrs=(size=12pt weight=bold) tickvalueattrs=(size=12pt weight=bold)) yaxisopts=(offsetmin=0.05 offsetmax=0.05 label="Results" labelfitpolicy=splitalways labelsplitchar='^' labelattrs=(size=11pt weight=bold) tickvalueattrs=(size=12pt weight=bold) linearopts=(tickvaluesequence=(start=-2000 end=15000 increment=1000) viewmin=-2000 viewmax=15000)); /* Reference Line */ referenceline y=0 / lineattrs=(color=grey pattern=3); /* Box Plot */ boxplot x=agegrpn y=AVAL / group=trtn groupdisplay=cluster boxwidth=0.56 clusterwidth=0.53 ; /*scatter of mean*/ scatterplot x=agegrpn y=mean1/group=treatment groupdisplay=cluster clusterwidth=0.53 markerattrs=(symbol=trianglefilled color=blue); scatterplot x=agegrpn y=mean2/group=treatment groupdisplay=cluster clusterwidth=0.53 markerattrs=(symbol=squarefilled color=green); scatterplot x=agegrpn y=mean3/group=treatment groupdisplay=cluster clusterwidth=0.53 markerattrs=(symbol=circlefilled color=grey); /*scatter of median*/ highlowplot x=agegrpn low=median1 high=median1/group=treatment groupdisplay=cluster clusterwidth=0.53 type=bar barwidth=0.6 outlineattrs=(thickness=2px color=blue) ; highlowplot x=agegrpn low=median2 high=median2/group=treatment groupdisplay=cluster clusterwidth=0.53 type=bar barwidth=0.6 outlineattrs=(thickness=2px color=green) ; highlowplot x=agegrpn low=median3 high=median3/group=treatment groupdisplay=cluster clusterwidth=0.53 type=bar barwidth=0.6 outlineattrs=(thickness=2px color=grey) ; /* Legend */ discretelegend 'A' 'B' 'R' / down=1 border=false; /* Axis Table */ innermargin; axistable x=agegrpn value=num / class=trtn classdisplay=cluster clusterwidth=0.53 classorder=ascending valueattrs=(weight=bold) labelattrs=(weight=bold) colorgroup=trtgrp ; endinnermargin; endlayout; endgraph; end; run; proc format; value agegrpn 1="5-13" 2="14-21" 3="22-35" ; value treatment 1="DrugA" 2="DrugB" 3="Refer" ; run; ods graphics/reset=all; proc sgrender data=final template=boxplot; format trtn treatment. agegrpn agegrpn.; run;
P.S.
You need to make a GROUP variable to make it happen if you want to use MERGEDLEGEND.
/***********************/ proc sort data=sashelp.class out =class; by sex; run; proc template; define statgraph mergedLegend; begingraph; entrytitle "Linear Regression By Gender"; layout overlay; scatterplot x=height y=weight / group=sex name="scat"; regressionplot x=height y=weight/ group=sex name="reg"; mergedlegend "scat" "reg" / border=true; endlayout; endgraph; end; proc sgrender data=class template=mergedLegend; run;
It's finally time to hack! Remember to visit the SAS Hacker's Hub regularly for news and updates.
Learn how use the CAT functions in SAS to join values from multiple variables into a single value.
Find more tutorials on the SAS Users YouTube channel.
Ready to level-up your skills? Choose your own adventure.