Skip to content

Reinforcement Rules

The reinforcement rules exploit a set of lever variables among the variables of a classifier to improve prediction scores.

ClassifierReinforcer

Structure(ClassifierReinforcer) ClassifierReinforcer(Structure(Classifier), 
                                                     Structure(VectorC) leverVariableNames)

Builds a ClassifierReinforcer structure from a Classifier structure and a list of lever variables. The resulting ClassifierReinforcer contains all the necessary information to compute reinforcement scores for each lever variable and target value.

ReinforcementInitialScoreAt

Numerical ReinforcementInitialScoreAt(Structure(ClassifierReinforcer), Categorical targetValue)

Returns the initial prediction score for the prediction of a given target value.

ReinforcementVariableAt

Categorical ContributionVariableAt(Structure(ClassifierReinforcer),
                                   Categorical targetValue, Numerical rank)

Returns the name of the lever variable at the specified reinforcement rank (starting at 1) for a target value, based on lever variables ordered by decreasing reinforcement scores.

ReinforcementPartAt

Categorical ReinforcementPartAt(Structure(ClassifierReinforcer),
                                Categorical targetValue, Numerical rank)

Returns the label of the lever variable part at the specified reinforcement rank (starting at 1) for a target value, based on lever variables ordered by decreasing reinforcement scores.

ReinforcementFinalScoreAt

Numerical ReinforcementFinalScoreAt(Structure(ClassifierReinforcer),
                                    Categorical targetValue, Numerical rank)

Returns the final score after reinforcement at the specified reinforcement rank (starting at 1) for a target value, based on lever variables ordered by decreasing reinforcement scores.

ReinforcementClassChangeTagAt

Numerical ReinforcementFinalScoreAt(Structure(ClassifierReinforcer),
                                    Categorical targetValue, Numerical rank)

Returns the class change tag after reinforcement at the specified reinforcement rank (starting at 1) for a target value, based on lever variables ordered by decreasing reinforcement scores:

  • 0 indicates the initial predicted target value was already the target to reinforce,

  • -1 indicates the final predicted target value remains different from the target to reinforce,

  • 1 indicates the final predicted target value is now the target to reinforce.