The Transform Variables node in SAS^® Enterprise Miner™ enables you to specify a Dummy Indicators transformation for class variables. However, if two or more of the class variables that are transformed have names that are identical in the first 24 characters, then the results of the transformation are incorrect. The resulting "TI_.." variables are incorrectly created with identical names. Because names are identical, only one of each name is kept.

There are no errors or warnings to indicate that a problem occurred.

To determine whether the problem occurs in any specific case, examine the Transformations Statistics in the node results. Look in the Variable Name column. For any names that are identical, only one of them is kept in the exported data.

To work around the problem, make sure that no variable names are identical in the first 24 characters.

EXAMPLE

A data set contains these two class variables:

The first 24 characters of the two variable names are not unique: "twenty_five_letter_var__".

In the Transform Variables node, choose the Default Methods Class Inputs property, Dummy Indicators. Run the Transform Variables node. Right-click the node, and choose Results. The Transformations Statistics pane shows these variables:

However, the exported data shows only one set.

Variable Name                       Label
TI_twenty_five_letter_var__1        twenty_five_letter_var__2:5
TI_twenty_five_letter_var__2        twenty_five_letter_var__2:6
TI_twenty_five_letter_var__3        twenty_five_letter_var__1:3
TI_twenty_five_letter_var__4        twenty_five_letter_var__1:4

Operating System and Release Information