Prediction Rules

An prediction rule has the form conditions -> conclusion (support, confidence) and expresses a relation of one or more items - the conditions - and one item as the conclusion. The quality of an association rule is expressed by its support and confidence measure.

The support of a prediction rule is the percentage of cases in which it applies in the training set. The confidence of a prediction rule is the percentage of cases in which the rule is correct relative to the number of cases in which it is applicable,\ie the conditions are met. An example prediction rule might be seg 2, seg 3 -> seg 1 (27\%, 93\%). So with the knowledge that plaques are present in seg 2 and seg 3 - which was true in 27% of all observations in the training data set - there is a probability of 93% that also a plaque in seg 1 is present.

More information about the computation of prediction rules is available in the corresponding publications.