Decision path
The decision path is displayed in textual form in the Decision path section. The score of the selected node is also displayed. It indicates the decision, this means, the prediction for the records that belong to the selected node.
The decision path lists properties for records in the selected node. Starting at the root node, a condition is shown for each node on the tree branch that connects the root and the selected node. Each record in the node must fulfill all the conditions on the decision path.
Typically, each condition restricts the value range in a data field.
- For categorical fields, the symbols =, ≠,
isIn, orisNotInmight occur in a condition. For example, a decision might look like this:relationship = Husband - For numeric fields, the symbols >, <, ≥, or ≤ might occur
in a condition. For example, the decision might look like this:
AGE ≥ 40
For each field type, the decision might also test whether a field
value is known or missing.
- A field value is known if the field contains any value but
NULL. For example, the condition might look like this:AGE IS NOT MISSING - A field value is missing if the field contains the
NULLvalue. For example, the condition might look like this:NUM_VESSELS IS MISSING
Independent of any field value, a decision can always be true (always fulfilled) or false (never
fulfilled). The corresponding conditions are True or False.
In complex situations, simple conditions can be joined to compound
formulas with the logical connectors
AND, OR,
or XOR. For example, compound formulas might look
like this: (relationship = husband) AND (AGE ≥ 40)
Conditions can also consist of a cascade. Cascades include an ordered
list of conditions. The first condition that can be evaluated is used.
An evaluation is possible if all the fields that occur in the condition
have known values. For example, a cascading decision might look like
this:
(NUM_VESSELS = 0) CASCADE ((AGE > 40) AND (ANGINA ≠ 1)) CASCADE TRUE
Instead of
CASCADE, the term SURROGATE can
also be used because the alternatives in the list are fall-back conditions.
If the term SURROGATE is used, the above example
looks like this: (NUM_VESSELS = 0) SURROGATE ((AGE > 40) AND (ANGINA ≠ 1)) SURROGATE TRUE