WaterFall Chart Dummy legend display if no data

[raja777pharma]

Hi,

I am trying to genrate the waterfall plot , but in data there is no all legend vlauee and only diaplay in legend what are in data.

But i need to display all values which are defined in proc format.

Below is sample output :

My Code is  :

ods listing close;
ods graphics on / border=off width=22.86cm height=9.5cm;

   proc sgplot data=final noborder noautolegend;
   by trt01a;
    format resp $rsf.;
    refline -50 / axis=y lineattrs=(pattern=shortdash);

	vbar subjid/response = pchg categoryorder=RESPDESC group=resp;

	xaxis label = "Patient Number" fitpolicy = ROTATE;
	yaxis label = "Maximum Percent Reduction from Baseline in SPD" min=&minp. max=&maxp. values=(-120 to 50 by 20);

   keylegend /title = 'IPCGS' location=inside position=topright;
run;

ods rtf close;
ods listing;
ods graphics off;

My current output : add one more group that is "PD" progressive disease 

How to display this water fall as per sample display 

Thank you.

Raja

[DanH_sas]

You can use a discrete attributes map to do it. You can use the map to control both the attributes for the plot , as well as whether the legend content is driven by the map instead of the data. This last behavior is controlled by the SHOW=ATTRMAP option in the map data. Here is a simple example of a map using the active style's attributes but using SHOW=ATTRMAP to have its content used for the legend. In your case, put all five possible responses in the map.

data attrmap;
    retain ID "myid" show "attrmap";
    length value $ 7 fillstyleelement $ 10 linestyleelement $ 10; 
    input value $ fillstyleelement $ linestyleelement $;
    cards;
    M       GraphData1 GraphData1
    F       GraphData2 GraphData2
    Unknown GraphData3 GraphData3
    ;
run;

proc sgplot data=sashelp.class dattrmap=attrmap;
    vbar age / response=weight group=sex groupdisplay=cluster stat=mean attrid=myid;
run;

[Ksharp]

Another workaround way is using LEGENDITEM statement.

proc sgplot data=sashelp.class;
    vbar age / response=weight group=sex groupdisplay=cluster stat=mean name='x';
	legenditem name='u' type=fill/fillattrs=(color=gray) label='Unknown';
	keylegend 'x' 'u';
run;