Adding Titles, Legends, and Insets: Adding Titles, Legends, and Insets

Adding Titles, Legends, and Insets

In Chapter 7, "Reading ODS Tables," you learned how to write ODS tables to SAS data sets and then how to read variables from the data sets into IML matrices. This chapter shows you how to create titles, legends, and insets on plots. (An inset is a box drawn on the plot containing statistics.) In particular, you learn how to incorporate the information from ODS tables into titles, legends, and insets.

The program statements in this chapter are distributed with Stat Studio. To open the program containing the statements:

Select File $\blacktriangleright\,$ Open $\blacktriangleright\,$ File from the main menu.
Click Go to Installation directory near the bottom of the dialog box.
Navigate to the Programs\Doc\STATGuide folder.
Select the Titles.sx file.
Click Open.

Suppose you perform a linear regression with PROC REG, and you create a scatter plot showing the fitted model and 95% confidence intervals for individual predictions as shown in Figure 8.1. You learned how to do this in Chapter 6, "Adding Curves to Plots."

Figure 8.1: A Fitted Model and Confidence Intervals

Suppose further that you want the plot to include the following information about the linear regression:

a title that shows the fitted model
a legend that explains the different colors of lines in the plot
an inset that shows the number of observations used in the regression and the R-square value of the fitted model

In Chapter 7, "Reading ODS Tables," you learned how to get all of the needed information out of the ODS tables output by using PROC REG. The program that follows combines ideas from previous chapters. Copy the following statements into a program window, and select Program $\blacktriangleright\,$ Run from the main menu.

  
 declare DataObject dobj; 
 dobj = DataObject.CreateFromFile("Hurricanes"); 
  
 dobj.WriteVarsToServerDataSet( {"wind_kts" "min_pressure"}, 
    "work", "Hurr", true ); 
  
 submit; 
 proc reg data=hurr; 
    model wind_kts = min_pressure; 
    output out=RegOut P=Pred LCL=LCL UCL=UCL; 
    ods output NObs = RegNObs 
               FitStatistics = RegFitStat 
               ParameterEstimates = RegParamEst; 
 run; 
 endsubmit; 
  
 ok = CopyServerDataToDataObject( "work", "RegOut", dobj, 
    {"Pred" "LCL" "UCL"}, /* names on server */ 
    {"Pred" "LCL" "UCL"}, /* names in DataObject */ 
    {"Predicted" "Lower Conf. Limit" "Upper Conf. Limit"}, 
    true ); 
  
 declare ScatterPlot FitPlot; 
 FitPlot = ScatterPlot.Create( dobj, "min_pressure", "wind_kts" ); 
  
 dobj.Sort( "min_pressure" ); 
 dobj.GetVarData( "min_pressure", minPress ); 
 dobj.GetVarData( "Pred", pred ); 
 FitPlot.DrawUseDataCoordinates(); 
 FitPlot.DrawSetPenColor( BLUE ); 
 FitPlot.DrawLine( minPress, pred ); 
  
 dobj.GetVarData( "LCL", lower ); 
 dobj.GetVarData( "UCL", upper ); 
 FitPlot.DrawSetPenColor( GRAY ); 
 FitPlot.DrawLine( minPress, lower ); 
 FitPlot.DrawLine( minPress, upper ); 
  
 use RegNObs;     read all var {N};        close RegNObs; 
 use RegFitStat;  read all var {NValue2};  close RegFitStat; 
 use RegParamEst; read all var {Estimate}; close RegParamEst; 
  
 print "--------------------------------------"; 
 print N[ rowname={"Num Read","Num Used","Num Missing"} ], 
       NValue2[ rowname={"R-Square","Adj R-sq"," "} ], 
       Estimate[ rowname={"Intercept","min_pressure"} ];

You can add a title to a plot by calling the SetTitleText and ShowTitle methods of the Plot class. However, in this case you first need to convert the values stored in the Estimate matrix from numerical to character values. One way to do this is to apply a w.d format by using the Base function PUTN. You can then concatenate pieces of the title together using the IML CONCAT function, and display the result.

The statements that follow build the title intelligently. Let be the intercept and be the slope of the regression line. The regression line for this example has negative slope ( $b_1 \lt 0$ ). If you built the title as

  
 title = concat("wind_kts = ", b0, " + ", b1, " * min_pressure");

then the title string might appear as

  
 wind_kts = 1333.35 + -1.29 * min_pressure

While this is correct, it is awkward to read. A more aesthetic title would be

  
 wind_kts = 1333.35 - 1.29 * min_pressure

This can be accomplished by treating the case of negative slope separately from the case of positive slope.

Add the following statements at the bottom of the program window, and select Program $\blacktriangleright\,$ Run from the main menu. Figure 8.2 shows how the title appears.

  
 b0 = putn(Estimate[1], "7.2"); 
 if Estimate[2]<0 then do;          /* if slope is negative */ 
    sign = " - ";                   /* display b0 - (-b1)   */ 
    b1 = putn(-Estimate[2], "4.2"); 
    end; 
 else do;                           /* else display b0 + b1 */ 
    sign = " + "; 
    b1 = putn(Estimate[2], "4.2"); 
    end; 
 title = concat("wind_kts = ", b0, sign, b1, " * min_pressure" ); 
 FitPlot.SetTitleText( title ); 
 FitPlot.ShowTitle();

Adding a legend is usually accomplished by using the DrawLegend module that is distributed with Stat Studio. The module is documented in the Stat Studio online Help . The statements that follow show one choice for displaying a legend. Add these at the bottom of the program window, and select Program $\blacktriangleright\,$ Run from the main menu. Figure 8.2 shows how the legend appears.

  
 Labels = {"Least-Squares Fit" "95% Prediction Limits"}; 
 LabelSize = 8;            /* 8 point font */ 
 LineColor = BLUE || GRAY; /* form 1x2 matrix */ 
 LineStyle = SOLID;        /* all lines are solid */ 
 Symbol = -1;              /* no symbols */ 
 BackgroundColor = WHITE; 
 Location = 'ILB';         /* Inside, Left, Bottom */ 
 run DrawLegend( FitPlot, Labels, LabelSize, 
                   LineColor, LineStyle, Symbol, 
                   BackgroundColor, Location );

Note the use of the IML concatenation operator ||. This operator takes the two predefined integers BLUE and GRAY and concatenates them into a matrix.

Caution: Trying to form the matrix by using the statement

  
 LineColor = {BLUE GRAY}; /* wrong! */

does not work because IML interprets that statement as forming a character matrix with values "BLUE" and "GRAY." Similarly, as the next statement illustrates, you must use the concatenation operator when creating a matrix from submatrices. You cannot write

  
 Values = {N[2] NValue2[1]}; /* wrong! */

You can add an inset by using the DrawInset module that is distributed with Stat Studio. The module is documented in the Stat Studio online Help . The code that follows shows one choice for creating an inset. Add the following statements at the bottom of the program window, and select Program $\blacktriangleright\,$ Run from the main menu. Figure 8.2 shows how the inset appears.

  
 Labels = {"Num Obs" "R-Square"}; 
 Values = N[2] || NValue2[1]; 
 LabelProps = .;          /* accept default settings for labels */ 
 LabelTypeface = "Arial"; /* font */ 
 BackgroundColor = -1;    /* no color (transparent) */ 
 Location = 'IRT';        /* Inside, Right, Top */ 
 run DrawInset( FitPlot, Labels, Values, 
                  LabelProps, LabelTypeface, 
                  BackgroundColor, Location );

Figure 8.2: Plot with Title,Legend,and Inset

Top of Page