createBranch(weight:attribute:)
Creates a child node that the decision tree should use as the result of a random choice, biased by the specified weight.
Declaration
func createBranch(weight: Int, attribute: any NSObjectProtocol) -> SelfParameters
- weight:
The relative probability that a random choice should follow this branch.
- attribute:
The attribute for the child node to create.
Return Value
The newly created child node.
Discussion
Use this method to specify a possible result of testing this node’s attribute (that is, an answer for this node’s question). The attribute parameter an have either of two kinds of values: if this test produces the decision tree’s final result (or action), the value identifies that action; if the next step in the decision tree is to be another question, the value identifies the next attribute to test (or question to ask).
This method adds a random branch to the decision tree; when you use the findActionForAnswers method, the tree tests the other attributes using the answers you specify in that method, but when it reaches an attribute with random branches, it chooses one randomly instead of testing the answer for that attribute. (To use random branching, you must set the randomSource property of the decision tree.) The weight parameter modifies the relative likelihood of each branch in a random set—branches with a higher weight are more likely than those with lower weights.
You can use this method to add some unpredictability to an otherwise deterministic decision; for example, a strategy combat game might use this method to choose an attack at random after other criteria have been satisfied. Here, the “Psychic Strike” attack will randomly occur 10% of the time (because its weight is 1 and the sum of all weights for the node it branches from is 10):