2D scatterplot with 3rd continuous var - color scale for missing

[Robert_Kozarski]

Dear SAS users,

How to make the grey-filled dots to be white-filled or 100% transparent (correspond to the Z missing):

 here comes the code:

data data_xyz;
   input X Y Z;
datalines;
1 5 10 
2 6 20 
3 7 .
4 8 40 
5 9 . 
6 10 60 
;
run;


proc template;
define statgraph scatter3Dcol;
dynamic valueX valueY valueZ;
begingraph;
	layout overlay / xaxisopts=(label='X' linearopts=(viewmin=0 viewmax=10)) 
					yaxisopts=(label='Y' linearopts=(viewmin=0 viewmax=10));
      scatterplot x=valueX y=valueY / markercolorgradient=valueZ colormodel=(CX3288BD CX99D594 CXE6F598 CXFEE08B CXFC8D59 CXD53E4F) 
      markerattrs=(symbol=circlefilled size=13 transparency=0.3) FILLEDOUTLINEDMARKERS=TRUE name="Scatter";
      continuouslegend "Scatter" / title='Observed X levels'; 
   endlayout;
endgraph;
end;
run;

proc sgrender data=data_xyz template=scatter3Dcol;
dynamic valueX="X" valueY="Y" valueZ ='Z';
run;

Thanks,

Robert

[Reeza]

Use a data attribute map. Not sure how that works with PROC TEMPLATE but it works well with SGPLOT.

[Robert_Kozarski]

Hi Reeza,

Thank you for your prompt response. For some reason I need to have it done under the proc template framework. I will try with attribute map.

Best,

Robert

[Rick_SAS]

Do you want the entire marker to be missing/transparent? If so, use a WHERE clause on the DATA= option:

proc sgrender data=data_xyz(where=(Z NE .)) template=scatter3Dcol; 

[Robert_Kozarski]

Thanks Rick,

I need the frames for my non-associated points. Ive just posted a solution with overwritten non-missing and missing Z observations.

Robert

[Ksharp · Accepted solution]

Override them ? by plotting these two points separately twice ?

[Robert_Kozarski]

Thanks for your responses. Ive overwritten the scatterplots with non-missing and missing Z.  Below is the code, even though a bit rough and ready.. If you think I could make it more neat pls do let me know.

Robert

data data_xyz;
   input X Y Z;
datalines;
1 5 10 
2 6 20 
3 7 .
4 8 40 
5 9 . 
6 10 60 
;
run;

data data_xyz1;
set data_xyz;
where Z ne .;
rename X=X1 Y=Y1 Z=Z1;
run;

data data_xyz2;
set data_xyz;
where Z eq .;
rename X=X2 Y=Y2;
run;

data data_xyz3;
merge data_xyz1 data_xyz2;
run;

proc print data= data_xyz3;run;

proc template;
define statgraph scatter3Dcol_1;
dynamic valueX1 valueY1 valueZ1 valueX2 valueY2 valueZ2 _trans _size;
begingraph;
	layout overlay / xaxisopts=(label='X' linearopts=(viewmin=0 viewmax=10)) 
					yaxisopts=(label='Y' linearopts=(viewmin=0 viewmax=10));
      scatterplot x=valueX1 y=valueY1 / markercolorgradient=valueZ1 colormodel=(CX3288BD CX99D594 CXE6F598 CXFEE08B CXFC8D59 CXD53E4F) 
      markerattrs=(symbol=circlefilled size=_size) datatransparency=_trans FILLEDOUTLINEDMARKERS=TRUE name="Scatter";
      scatterplot x=valueX2 y=valueY2 /
      markerattrs=(symbol=circle color=black size=_size) datatransparency=_trans FILLEDOUTLINEDMARKERS=TRUE name="Scatter1";
      continuouslegend "Scatter" / title='Observed Z levels'; 
   endlayout;
endgraph;
end;
run;

proc sgrender data=data_xyz3 template=scatter3Dcol_1;
dynamic valueX1="X1" valueY1="Y1" valueZ1 ='Z1' valueX2="X2" valueY2="Y2" _trans=0 _size=13;
run;

resulting with: