Solved: Re: PROC SGPLOT axis label creation and Legend Issue

SASuserlot · Posted 12-10-2021 10:24 AM

I am trying to create a graph, using SGPLOT with both scatter and series statement.

1. Please correct me if I am using the wrong type of Graphs statement for this kind of graphs ( I just want to show %chg Vs percentage of population). I want to show data point and a line connecting them.

2. Why my key legend shows 4 different groups, instead of showing two groups where circle should overlap the line?

3. I would like to insert a big arrow symbol on both sides of '0' indicating score improve or Decline!.

data trial;
input subject trt response chg;
cards;
1001 1 15 20
1002 1 16 25
1003 1 17 30
1004 1 -6 -20
1005 1 0 0
1006 1 -5 -25
1007 1 15 20
1008 1 16 25
1009 1 17 30
1010 1 18 30
1011 2 29 50
1012 2 -4 -12
1013 2 13 10
1014 2 12 8
1015 2 0 0
1016 2 28 45
1017 2 29 50
1018 2 -4 -12
1019 2 29 50
1020 2 -4 -12

;
run;

proc sort data=trial out= trial1; by trt response; run;
data trial2;
set trial1;
by trt response;
 if first.response then cnt=0;
 if .<chg<=response then cnt+1;
 if .<chg>response then cnt+1;
  if last.response then do;
   if trt=1 then yvalue=cnt/10 ;*10- indicated total poputation where trt=1 and 2;
   else if trt=2 then yvalue=cnt/10;
  end;
 run;

ods listing close;                                                                                                                      
ods html ;*file='scatterplot.html' path='.' style=styles.statistical;                                                                     
ods graphics / reset width=600px height=400px imagefmt=gif;                                                                             
                                                                                                                                        
/* The SCATTER statement generates the scatter plot with error bars. */                                                                 
/* The SERIES statement draws the line to connect the means.         */                                                                 
proc sgplot data=trial2 noautolegend;                                                                                                  
   scatter x=chg y=yvalue / group= trt;* markerattrs=(color=blue symbol=CircleFilled);                                                                
   series x=chg y=yvalue / group= trt ;*lineattrs=(color=blue pattern=2);  
	xaxis  min=-40 max=40; 
	yaxis  min=0 max= 0.5; 
keylegend / valueattrs=(size=12pt) titleattrs=(size=14pt);
run;                                                                                                                                    
                                                                                                                                        
ods html close;

Ksharp · Posted 12-11-2021 06:31 AM

data trial;
input subject trt response chg;
cards;
1001 1 15 20
1002 1 16 25
1003 1 17 30
1004 1 -6 -20
1005 1 0 0
1006 1 -5 -25
1007 1 15 20
1008 1 16 25
1009 1 17 30
1010 1 18 30
1011 2 29 50
1012 2 -4 -12
1013 2 13 10
1014 2 12 8
1015 2 0 0
1016 2 28 45
1017 2 29 50
1018 2 -4 -12
1019 2 29 50
1020 2 -4 -12

;
run;

proc sort data=trial out= trial1; by trt response; run;
data trial2;
set trial1;
by trt response;
 if first.response then cnt=0;
 if .<chg<=response then cnt+1;
 if .<chg>response then cnt+1;
  if last.response then do;
   if trt=1 then yvalue=cnt/10 ;*10- indicated total poputation where trt=1 and 2;
   else if trt=2 then yvalue=cnt/10;
  end;
 run;

data arrow;
y=0;y1=0.05;
low1=-30;high1=-10;x1=-20;text='YYYYYYYY';output;
call missing(of _all_);
y=0;y1=0.05;
low2=10 ;high2=30;x2=20;text='XXXXXXX';output;
run;
data trial3;
 set trial2 arrow;
run;

ods graphics / reset width=600px height=400px imagefmt=gif;                                                                             
                                                                                                                                        
/* The SCATTER statement generates the scatter plot with error bars. */                                                                 
/* The SERIES statement draws the line to connect the means.         */                                                                 
proc sgplot data=trial3 ;                                                                                                  
   series x=chg y=yvalue / group= trt markers name='a';*lineattrs=(color=blue pattern=2);  
   highlow y=y low=low1 high=high1/lowcap=BARBEDARROW ; 
   highlow y=y low=low2 high=high2/highcap=BARBEDARROW ;
   text y=y1 x=x1 text=text/strip CONTRIBUTEOFFSETS=none;
   text y=y1 x=x2 text=text/strip CONTRIBUTEOFFSETS=none;
	xaxis  min=-40 max=40; 
	yaxis  min=0 max= 0.5; 
keylegend 'a' / valueattrs=(size=12pt) titleattrs=(size=14pt);
run;

View solution in original post

ballardw · Posted 12-10-2021 10:57 AM

Remove the scatter statement; add the options MARKERS and MARKERATTRS=(SYMBOL=CIRCLE) to the series statement.

That way you only have 1 legend element and you can't get different legends to overlay.

There are many other other Marker related options for the Series statement so feel free to play with them.

@SASuserlot wrote:

I am trying to create a graph, using SGPLOT with both scatter and series statement.

1. Please correct me if I am using the wrong type of Graphs statement for this kind of graphs ( I just want to show %chg Vs percentage of population). I want to show data point and a line connecting them.

2. Why my key legend shows 4 different groups, instead of showing two groups where circle should overlap the line?

3. I would like to insert a big arrow symbol on both sides of '0' indicating score improve or Decline!.

data trial;
input subject trt response chg;
cards;
1001 1 15 20
1002 1 16 25
1003 1 17 30
1004 1 -6 -20
1005 1 0 0
1006 1 -5 -25
1007 1 15 20
1008 1 16 25
1009 1 17 30
1010 1 18 30
1011 2 29 50
1012 2 -4 -12
1013 2 13 10
1014 2 12 8
1015 2 0 0
1016 2 28 45
1017 2 29 50
1018 2 -4 -12
1019 2 29 50
1020 2 -4 -12

;
run;

proc sort data=trial out= trial1; by trt response; run;
data trial2;
set trial1;
by trt response;
 if first.response then cnt=0;
 if .<chg<=response then cnt+1;
 if .<chg>response then cnt+1;
  if last.response then do;
   if trt=1 then yvalue=cnt/10 ;*10- indicated total poputation where trt=1 and 2;
   else if trt=2 then yvalue=cnt/10;
  end;
 run;

ods listing close;                                                                                                                      
ods html ;*file='scatterplot.html' path='.' style=styles.statistical;                                                                     
ods graphics / reset width=600px height=400px imagefmt=gif;                                                                             
                                                                                                                                        
/* The SCATTER statement generates the scatter plot with error bars. */                                                                 
/* The SERIES statement draws the line to connect the means.         */                                                                 
proc sgplot data=trial2 noautolegend;                                                                                                  
   scatter x=chg y=yvalue / group= trt;* markerattrs=(color=blue symbol=CircleFilled);                                                                
   series x=chg y=yvalue / group= trt ;*lineattrs=(color=blue pattern=2);  
	xaxis  min=-40 max=40; 
	yaxis  min=0 max= 0.5; 
keylegend / valueattrs=(size=12pt) titleattrs=(size=14pt);
run;                                                                                                                                    
                                                                                                                                        
ods html close;

SASuserlot · Posted 12-10-2021 12:49 PM

Legend, data points fixed. Thanks @ballardw . How about inserting the 'Arrow' Labels under the axis like I mentioned in the image.

Ksharp · Posted 12-11-2021 06:11 AM

https://blogs.sas.com/content/graphicallyspeaking/2016/05/30/ctspedia-clinical-graphs-subgrouped-for...

https://blogs.sas.com/content/graphicallyspeaking/2012/09/30/forest-plot-with-subgroups/

SASuserlot · Posted 12-11-2021 12:13 PM

Thank you very much.

Ksharp · Posted 12-11-2021 06:31 AM

data trial;
input subject trt response chg;
cards;
1001 1 15 20
1002 1 16 25
1003 1 17 30
1004 1 -6 -20
1005 1 0 0
1006 1 -5 -25
1007 1 15 20
1008 1 16 25
1009 1 17 30
1010 1 18 30
1011 2 29 50
1012 2 -4 -12
1013 2 13 10
1014 2 12 8
1015 2 0 0
1016 2 28 45
1017 2 29 50
1018 2 -4 -12
1019 2 29 50
1020 2 -4 -12

;
run;

proc sort data=trial out= trial1; by trt response; run;
data trial2;
set trial1;
by trt response;
 if first.response then cnt=0;
 if .<chg<=response then cnt+1;
 if .<chg>response then cnt+1;
  if last.response then do;
   if trt=1 then yvalue=cnt/10 ;*10- indicated total poputation where trt=1 and 2;
   else if trt=2 then yvalue=cnt/10;
  end;
 run;

data arrow;
y=0;y1=0.05;
low1=-30;high1=-10;x1=-20;text='YYYYYYYY';output;
call missing(of _all_);
y=0;y1=0.05;
low2=10 ;high2=30;x2=20;text='XXXXXXX';output;
run;
data trial3;
 set trial2 arrow;
run;

ods graphics / reset width=600px height=400px imagefmt=gif;                                                                             
                                                                                                                                        
/* The SCATTER statement generates the scatter plot with error bars. */                                                                 
/* The SERIES statement draws the line to connect the means.         */                                                                 
proc sgplot data=trial3 ;                                                                                                  
   series x=chg y=yvalue / group= trt markers name='a';*lineattrs=(color=blue pattern=2);  
   highlow y=y low=low1 high=high1/lowcap=BARBEDARROW ; 
   highlow y=y low=low2 high=high2/highcap=BARBEDARROW ;
   text y=y1 x=x1 text=text/strip CONTRIBUTEOFFSETS=none;
   text y=y1 x=x2 text=text/strip CONTRIBUTEOFFSETS=none;
	xaxis  min=-40 max=40; 
	yaxis  min=0 max= 0.5; 
keylegend 'a' / valueattrs=(size=12pt) titleattrs=(size=14pt);
run;

SASuserlot · Posted 12-11-2021 11:42 AM

Perfect. for my future understanding .. we have to create a separate data sets guiding the axis points and append to the original dataset, in order to place any thing under the graph for the things that we can not be achieved in the sgplot? Thank you.

Ksharp · Posted 12-12-2021 04:11 AM

Yes. I think so.
Or you could refer to the blog URL I posted before.

SAS Innovate 2025: Save the Date

SAS Training: Just a Click Away