Language Reference

CALL HEATMAPCONT (x <*>COLORRAMP=ColorRamp <*>SCALE=scale <*>XVALUES=xValues<*>YVALUES=yValues<*>XAXISTOP=top<*>DISPLAYOUTLINES=outlines<*>TITLE=plotTitle<*>LEGENDTITLE=legendTitle<*>LEGENDLOC=loc <*>SHOWLEGEND=show <*>RANGE=range <*>OTHER=other );

The HEATMAPCONT subroutine is part of the IMLMLIB library . The HEATMAPCONT subroutine displays a heat map of a numeric matrix whose values are assumed to vary continuously. The heat map is produced by calling the SGRENDER procedure to render a template, which is created at run time. The argument x is a matrix that contains numeric data. The ODS statistical graphics subroutines are described in Chapter 16: Statistical Graphics.

A simple example follows. The numeric variables from the Sashelp.Cars data set are read into a matrix and the CORR function is used to compute the correlation matrix for those variables. The HEATMAPCONT subroutine creates the image in Figure 25.156, which visualizes the correlations. The correlation matrix has high values (1) on the main diagonal. There are large negative correlations between horsepower and the fule efficiency variables, MPG_City and MPG_Highway.

use Sashelp.Cars;
read all var _NUM_ into Y[c=varNames];
close Sashelp.Cars;
corr = corr(Y);
call HeatmapCont(corr) xvalues=varNames yvalues=varNames;

Specify the x vector inside parentheses and specify all options outside the parentheses. Titles are specified by using the TITLE= option. Each option corresponds to a statement or option in the graph template languge (GTL).

The following list documents the options to the HEATMAPCONT routine:

The following example shows how to create a heat map that uses the SCALE=, XVALUES=, YVALUES=, and TITLE= options.

use Sashelp.Class;
read all var _NUM_ into Students[c=varNames r=Name];
close Sashelp.Class;

/* sort data in descending order according to Age and Height */
call sortndx(idx, Students, 1:2, 1:2);
Students = Students[idx,];
Name = Name[idx];

/* standardize each column */
call HeatmapCont(Students) scale="Col"
        xvalues=varNames yvalues=Name title="Student Data";

In Figure 25.157, you can see that Philip is the biggest student, Joyce is the smallest, Robert is heavy for his age, and Alfred is tall for his age.

For a more complicated visualization of a data matrix, the following statements visualize the number of snack items solds at a fictitious store over the course of 1,022 days. The heat map that uses the YVALUES=, DISPLAYOUTLINES=, and TITLE= options. Because the quantity of items sold range over two orders of magnitude (from 0 to 121), a logarithmic transformation is used to transform the data.

use Sashelp.Snacks;
read all var {QtySold Date Product};
close Sashelp.Snacks;

QtySold = choose(QtySold>=0, QtySold, .); /* remove invalid quantities */
Names = unique(Product);
x = shape(QtySold, ncol(Names));

ods graphics / height=800 width=1400;
call HeatmapCont(log10(x+1)) yvalues=Names displayoutlines=0
                             title="Log10(Items Sold) by Day";

In Figure 25.158, horizontal white bands indicate periods of time for which a particular snack item was not sold. Vertical white bands indicate days for which the store was closed. Dark shades, such as for "classic potato chips" and "tortilla chips," indicate items for which the average number of units sold each day was about . Lighter shades, such as for "fiesta sticks" and "stone-ground wheat sticks," indicate less popular items.