SAS 9.4 Graph Template Language, 3rd Edition

can nested layouts. For information about adding a legend or other information

outside of the cells, see “Example 3: A Data Panel with Sidebars” on page 268.

• Axis options for classification panels are specified on the LAYOUT

DATALATTICE or LAYOUT DATAPANEL statement, not on the LAYOUT

PROTOTYPE statement. For information about setting axis options for the layout,

see “Managing Axes in DATALATTICE and DATAPANEL Layouts” on page 239.

Distinction between DATAPANEL and

DATALATTICE

The DATAPANEL and DATALATTICE layouts differ in how their classification

variables are declared and in how they populate their cells. For the DATAPANEL

layout, the classification variables are declared as a list of variables in parentheses in the

CLASSVAR= option as shown in the following example.

layout datapanel classvars=(product division) / ...;

The number of class variables in the list is unlimited, though the effectiveness of the

graph decreases as the number of class variables exceeds three or four. In such a case, it

is better to use two class variables, and use the other class variables in the BY statement

of the SGRENDER procedure.

For the DATALATTICE layout, one classification variable is assigned for a row

dimension, and one classification variable is assigned for a column dimension. The row

variable is assigned in the ROWVAR= option, and the column variable is assigned in the

COLUMNVAR= option as shown in the following example.

layout datalattice rowvar=product columnvar=division / ...;

Another key difference between the DATAPANEL and the DATALATTICE layouts is

how they create their cells. The DATAPANEL layout creates a cell for each

classification-variable crossing that produces data. Any crossing that does not produce

data is not included in the panel. Conversely, the DATALATTICE layout creates a cell

for each crossing of its classification variables regardless of whether it produce data. A

crossing the produces no data appears as an empty cell in the lattice. This difference is

significant when analyzing data that is sparse. In that case, the DATAPANEL layout

produces a panel that contains only cells that display data while the DATALATTICE

layout produces a lattice that might contain a large number of empty cells.

Organizing Panel Contents

Overview of What to Consider When Planning a Classification Panel

When planning a classification panel, the following factors influence the layout

specification:

• grid dimensions (number of rows and columns)

• cell population order as the layout is rendered

• gutters between the cells

• graph aspect ratio

Organizing Panel Contents 231

• cell size within the panel

• prototype orientation

Grid Dimensions and Cell Population Order

Assume you want to create a DATAPANEL layout with one classification variable that

has five unique values. Before starting to write code, you must first decide what grid

dimensions you want to set (how many columns and rows) and whether you want to

permit empty cells in the grid. If do not want empty cells, you must limit the grid to five

cells, which gives you two choices for the grid dimensions: five columns by one row

(5x1), or one column by five rows (1x5). If you are willing to have empty cells in the

grid, you could have several grid sizes, such as a 2x3 or a 3x2 grid.

The easiest way to specify a grid dimension is to set both the COLUMNS= and ROWS=

options to the desired number of columns and rows. If one dimension is set, the other

dimension automatically grows to accommodate the number of classification levels. By

default, COLUMNS=1, and the ROWS= option is not set.

By default, the layout uses the ORDER=ROWMAJOR setting to populate grid cells.

This specification essentially means "fill in all cells in the top row (starting at the top

left) and then continue to the next row below." The following layout leaves the default

ORDER=ROWMAJOR setting in effect:

layout datapanel classvars=(var) / columns=3 rows=2;

layout prototype;

... plot statements ...

endlayout;

endlayout;

Here is the resulting grid layout.

Alternatively, you can specify ORDER=COLUMNMAJOR, which populates the grid by

filling in all cells in the left column (starting at the top), and then continuing with the

next column:

layout datapanel classvars=(var) / columns=3 rows=2 order=columnmajor;

layout prototype;

... plot statements ...

endlayout;

endlayout;

232 Chapter 13 • Creating Classification Panels Using the DATALATTICE and DATAPANEL Layouts