CODEWITHSUNDEEP
graphsas
pyll
pyll

Reputation: 1764

Annotate Syntax PROC SGRENDER SAS

I have a dataset I am working with that is very similar to the one created below. I want to graph the yearly trend in score by state, on the same graph. Additionally, I want to shade a portion of the graph where something happened (in this case a drop in score). Let's say I expected this event to be between 2011 and 2013. I want to shade a light gray rectangle over those years. I think my code is close, but cannot quite get the syntax correct.

Any help accomplishing this would be greatly appreciated. Also, any recommendations on cleaning up the proc template statements (are both needed? or can I combine into one?). Thanks.

Edit: the main issue, I think, is the syntax of the annorec data step.

 data statescores;
        input state $ year score;
        cards;
    NC 2010 75
    NC 2011 77
    NC 2012 72
    NC 2013 85
    NC 2014 87
    SC 2010 72
    SC 2011 73
    SC 2012 60
    SC 2013 79
    SC 2014 76
    VA 2010 80
    VA 2011 80
    VA 2012 79
    VA 2013 81
    VA 2014 83
    ;
    run;

data drop;                                                                                                                        
   input startdate enddate;                                                                                             
   datalines;                                                                                                                           
2011 2013                                                                                                                        
;   

data annorec;
/*   length function style color $8;       */
    retain xsys '2' ysys '1' when 'b';                                                                                                    
    set drop;                                                                                                       
    function='rectangle';
    height=.4;
    width=2; 
    x=startdate;                                                                                                                        
    y=60;  
    display="all";
    FILLCOLOR='LTGRAY'; 
    output;                                                                                                           
/*   function='bar';             */
/*   x=enddate;                  */
/*   y=90;                     */
/*   color='ltgray';           */
/*   style='solid';             */
/*   output;                      */
run;     

proc template;
    define style styles.stocks;
    parent=styles.listing;
    style GraphData1 from GraphData1 /
        ContrastColor=blue
        Color=red
        MarkerSymbol="CircleFilled"
        Linestyle=1;
    style GraphData2 from GraphData2 /
        ContrastColor=brown
        Color=blue
        MarkerSymbol="none"
        Linestyle=2;
    style GraphData3 from GraphData3 /
        ContrastColor=orange
        Color=orange
        MarkerSymbol="none"
        Linestyle=2;
end;
run;

proc template;
    define statgraph trend;
    begingraph;
    entrytitle "Scores for NC, SC, and VA";
    discreteattrmap name="stockname" / ignorecase=true;
    value 'NC' /
    markerattrs=GraphData1(color=red symbol=circlefilled)
    lineattrs=GraphData1(color=red pattern=solid);
    value 'SC' /
    markerattrs=GraphData2(color=orange)
    lineattrs=GraphData2(color=green pattern=dot);
    value 'VA' /
    markerattrs=GraphData3(color=blue)
    lineattrs=GraphData3(color=blue pattern=dot);
    enddiscreteattrmap;
    discreteattrvar attrvar=stockmarkers var=stock
    attrmap="stockname";
    layout overlay;
    seriesplot x=year y=score /
    group=state
    display=(markers)
    name="scores";
    discretelegend "scores" / title="State Scores";
    endlayout;
    endgraph;
    end;
run;
/* Plot the score trends */
ods pdf file="/sasdata/username/testgraph.pdf";
ods pdf style=styles.stocks;


proc sgrender data=statescores template=trend sganno=annorec;
run;
quit;

ods pdf close;

Upvotes: 0

Views: 991

Answers (1)

Joe
Joe

Reputation: 63424

I would try to avoid using a discrete attribute map in GTL if you can avoid it. If you have only a few states, you really don't need it; even in your code above it does little other than to override some things you set in the earlier template step. Attribute maps are nice for SGPLOT, but are a pain in GTL in my experience.

In this case, you should try to use the BAND plot to make your gray band. Play with it some, but this should basically give you what you want:

proc sgplot data=statescores;
series x=year y=score/ group=state;
band y=score upper=2013 lower=2011 /transparency=0.8 ;
run;

That makes a pretty nice looking plot with four lines of code, plus if you want to add the colors you can fairly easily (either in an attribute map dataset, or in template code). In general, annotate datasets are often unnecessary in the SGPlot/GTL world, as you can layer plots easily and a lot of annotation can be done via layered plots instead.

If you want more detailed code review, I suggest posting in communities.sas.com, where some of the graph developers who work at SAS will often give suggestions on improvements to code.

Upvotes: 1

Related Questions