| Instances/Instance Properties | Decision Trees | |
Cases are split into two sections. The first, identifies what tied columns there are in the cases. The second, fills the cases. To specify the columns in the cases you need a <CASECOLUMN> tag for each column, followed by a tiedObject attribute name which is the name of an existing object, and a type attribute name which is the type of the object (List, Numeric, Boolean, Date). You do not need to specify the outcome column of the cases. Next, you can specify whether or not you want the cases to be cleared before filling in the new ones with the <CLEAR_CASES> tag, omitting this will append any cases you specify to any existing ones.
A case consists of a <CASE> tag followed by attribute names which correspond to the tiedObject names that you specified above followed by what value you want for that tiedObject (<tiedObject name>="<tiedObject value>"). The last attribute is the outcome of the case, use the value attribute name. The above example has three tied objects declared with the <CASECOLUMN> tag
...
<CASECOLUMN tiedObject="Grade" TYPE="LIST"/>
<CASECOLUMN tiedObject="Department" TYPE="LIST"/>
<CASECOLUMN tiedObject="Hotel" TYPE="LIST"/>
...
so a <CASE> line would look something like this:
...
<CASE Grade="Director" Department="" Hotel="" value="Pass"/>
...Notice Grade, Department, and Hotel become attribute names. Also Department and Hotel have no string value associated with them, this is the equivalent of "Don't Care" or "*" in a case.