DESCRIPTION

For example, if a scatterplot is being created with various sized points, there is no automatic way to get a legend of the various point sizes, so proc legendentry could be executed (probably several times) to define legend information, then proc legend could be executed to render it.

proc legendentry is also useful in setting up legend-driven plots which allow things like bar color, or scatterplot symbol color, size, or shape to be controlled by a data field.

EXAMPLE

PREREQUISITES

MANDATORY ATTRIBUTES

ATTRIBUTES

sampletype t

The type of sample to be displayed with the entry. t may be one of: line, color , symbol , text , line+symbol , or text+symbol, or none.
none may be used to insert spacing or text headers/separators into the legend.
Example: sampletype: text

label text

The label for the entry. Example: label: Eastern counties

tag string

An optional short identifier to be associated with the entry, so that the entry can be accessed later. Used with proc bars colorfield, as in this example: colorfld

details text

Details regarding the rendering of the sample. What this is depends on the sampletype.

For line, a linedetails specification should be given.

For color, a color should be given.

For symbol, a symboldetails specification should be given.

For text, the character or word should be given here, and a textdetails specification may be given using the details2 attribute.

For line+symbol, a linedetails specification should be given here, and a symboldetails specification given using the details2 attribute.

For text+symbol, the character or word should be given here, a textdetails specification should be given using the details2 attribute, and a symboldetails specification given using the details3 attribute.

If sampletype is none, nothing should be given for this attribute.

details2 text

Secondary details attribute. See details above.

details3 text

See details above.

LEGEND-DRIVEN TECHNIQUE

"Legend-driven technique" allows bar color, scatterplot symbol, etc. to be controlled by a data field. This technique involves defining a legend first, containing information to associate data field contents with appearance details. Then the plotting is done.

Here is the procedure:

1. For every expected variation, set up an explicit legend entry using proc legendentry. label, tag, and details attributes should be supplied. The tag will be compared with data field contents. The details attributes control appearance.

2. Plot, using the appropriate attribute. For example, if you are doing a bar graph where individual bar colors are controlled by a data field, you would use proc bars and specify a colorfield.

Examples of legend-driven technique are colorfld symfld symfld dupsleg