Help with legend

[VijayKumar]

If I fill some area on a line graph (that is using annotate dataset, I create a box in some area of the graph and color the box with say blue ).

Now how do I create a legend value of this color??  (the line graph will already be having a legend value associated with it  for say 'amount' variable or any statistic variable.)... so how do I add the above color as a legend  value to the existing default legend..???

[VijayKumar]

Do we have any experts to answer this question? Any suggestions will be highly appreciative...

[DanH_sas]

Here is how you can do it using the Graph Template Language (GTL) in SAS 9.3:

proc template;
define statgraph plot;
begingraph / drawspace=datavalue;
legenditem type=fill name="box" / label="The Alfred Box" fillattrs=(color=orange transparency=0.75); 
layout overlay;
  scatterplot y=weight x=height / group=sex name="scatter";
  drawrectangle x=69 y=112.5 width=10 height=15 / 
               display=all fillattrs=(color=orange) 
               transparency=0.75;  
  drawtext "Alfred" / x=69 y=112  anchor=top; 
  discretelegend "scatter" "box";
endlayout;
endgraph;
end;
run;

proc sgrender data=sashelp.class template=plot; run;

[VijayKumar]

Thanks DanH for your reply and providing a solution.

I am actually using SAS9.2.  Running your code gave error at drawspace and legenditem............

Still I tried running the below simple proc template via EG to check step by step..   but got error:
ERROR: Insufficient authorization to access ...path../SASApp/SGRender1.png.

proc template;
  define statgraph histogram;
    begingraph;
      layout overlay;
        histogram weight;
      endlayout;
    endgraph;
  end;
run;

ods graphics / width=450px;
ods listing;
proc sgrender data=sashelp.class
  template=histogram;
run;

Then I tried running same via unix prompt and got error:

ERROR: The connection to the X display server could not be made....

But when running normal gplot and gchart procedures its not giving X display error....

[GraphGuy]

Do you have an X-server running somewhere that you could set your 'DISPLAY' to?

(Maybe an XMing or Hummingbird eXceed X-server running somewhere?)

Such as:    export DISPLAY=somemachine.foo.unx:0.0

In the past, SAS/Graph running on Unix needed an X-display to query what fonts were available.

After SAS version 9.2, I don't think SAS/Graph needs an X-display anymore, but maybe the ODS Graphics (such as proc sgrender) need it(?) - I'm not 100% sure on that one, but it's probably worth a try.

[DanH_sas]

ODS Graphics does not need a display to run. When you say, "via unix prompt", are you running in batch, or with -nodms?

[VijayKumar]

On unix prompt ; I just type  >  sas program_name.sas

But when I run any other sas program that has GPLOT & GCHART  (via unix prompt or EG), it runs fine without any issues.

Before my previous reply , I had tried      export display=host:port  but that wasn't working...

[VijayKumar]

Below is the code that I can run in SAS9.2 and get the output. Is it possible to do some changes /addition to this code and get a legend displayed for the orange color in the graph.
I mean there must be some trick to fool SAS.. 

data carea;
LENGTH FUNCTION $8 STYLE $8 COLOR $8;
retain xsys ysys "2" hsys "3" when "b";
function = 'move'; x=2008; y=2000; output;
function="bar";  x=2011; y=6000; color="orange"; line=1; style="simplex"; output;
run; 
symbol v=dot i=join;
legend1 label=none;
proc gplot data=sales;
plot (sale) * year/ legend=legend1 annotate=carea;
run;
quit;

[VijayKumar]

Here is the output :

[GraphGuy]

The following example shows how to manually annotate a legend for the orange color:

%let name=anno_legend;
filename odsout '.';

data sales;
format Sale comma10.0;
input Year Sale;
datalines;
2002 2100
2003 4000
2004 3200
2005 5100
2006 3300
2007 2600
2008 5000
2009 5200
2010 5400
2011 5600
;
run;

data color_area;
LENGTH FUNCTION $8 STYLE $8 COLOR $8;

xsys='2'; ysys='2'; hsys='3'; when='b';
function = 'move'; x=2008; y=2000; output;
function="bar";  x=2011; y=6000;
color="orange"; line=1; style="solid"; output;

/* Draw a custom legend with annotate */
xsys='3'; ysys='3'; hsys='3'; when='a';
function = 'move'; x=60; y=1; output;
function="bar";  x=x+3; y=y+3;
color="orange"; line=1; style="solid"; output;
function="label"; color='black'; position='6'; x=64.5; y=3;
text="Orange Area"; output;

run;

goptions device=png;
ODS LISTING CLOSE;
ODS HTML path=odsout body="&name..htm" style=htmlblue;

symbol1 value=dot height=1.5 interpol=join color=black;
legend1 label=none;

axis1 minor=none offset=(0,0);
axis2 minor=none offset=(0,0);

proc gplot data=sales;
plot sale*year=1 /
legend=legend1 annotate=color_area
vaxis=axis1 haxis=axis2
des='' name="&name";
run;

quit;

ODS HTML CLOSE;
ODS LISTING;

[GraphGuy]

Here is the output:

[VijayKumar]

Thanks Robert !  This works...

The only thing is if footnotes are added or if the height of graph area changes, then adjustment will have to be made to the co-ordinates of this legend...right?   Other than that this looks great!!

One small question..

In above example to draw a bar , we start from x=2008, y=2000  till x1=2011 , y1=6000.

Suppose if the data keeps on changing then we dont know what exactly will be thevalue for y,x1,y1.
But we know only x (i.e., 2008...from where the fill area should start ) and that starting from 2008 till the maximum point on X-axis all of the area needs to be filled with a color. Is it possible to do something like .... Move to point 2008 on X-axis; then draw a straight line till the top of the plot area; then fill all of the area on the right side of this straight line?

[GraphGuy · Accepted solution]

In that case, you could use the '1' coordinate system (such that you can specify x/y coordinates in the range of 0-100% of the area inside the axes).  Here is the code that draws the orange area that way:

hsys='3'; when='b';

function = 'move'; xsys='2'; x=2008; ysys='1'; y=0; output;

function="bar";  xsys='1'; x=100; ysys='1'; y=100;

color="orange"; line=1; style="solid"; output;

[VijayKumar]

Excellent! this works.... Thanks Robert, u r great!!