Working with ODS Graphics

Graphics are an indispensable part of statistical analysis. Graphics reveal patterns, identify differences, and provoke meaningful questions about your data. Graphics add clarity to an analytical presentation and stimulate deeper investigation.

SAS 9.2 introduced the Graph Template Language (GTL), a powerful new language for defining clear and effective statistical graphics. The GTL enables you to generate various types of plots, such as model fit plots, distribution plots, comparative plots, prediction plots, and more.

The GTL applies accepted principles of graphics design to produce plots that are clean and uncluttered. Colors, fonts, and relative sizes of graph elements are designed for optimal impact. By default, the GTL produces PNG files, which support true color (the full 24-bit RGB color model) and enable visual effects such as anti-aliasing and transparency. A PNG file retains a small file size. GTL statement options enable you to control the content and appearance of the graphic down to the smallest detail.

The GTL is designed to produce graphics with minimal syntax. The GTL uses a flexible, building-block approach to create a graph by combining statements in a StatGraph template. StatGraph templates are defined with the TEMPLATE procedure.

With just a few statements, you can create the graphs that you need to analyze your data. For example, you can create the following model fit plot with these statements:

proc template;
define statgraph mytemplate;
beginGraph;
   entrytitle "Model Weight by Height";
   layout overlay;
   bandplot x=height limitupper=upper limitlower=lower;
   scatterplot y=weight x=height;
   seriesplot y=predict x=height;
endlayout;
endGraph;
end;
run;

proc sgrender data=sashelp.classfit
              template=mytemplate;
run;

The previous example defines a StatGraph template named MyTemplate, which uses values from the data set Sashelp.Classfit. This data set contains the data variables HEIGHT and WEIGHT and precomputed values for the fitted model (Y=PREDICT) and confidence band (limitupper=upper limitlower=lower). The SGRENDER procedure uses the data in Sashelp.Classfit and the MyTemplate template to render the graph. (This example is member GTLMFIT1 in the SAS Sample Library.)

The following two graphics are more examples of what you can do with ODS graphics: