Improving the Classifier CBA with an Additional Pruning Method

In data mining, classification based on association is a technique that employs association rule mining algorithms to construct a classifier by selecting a subset of the rules extracted from a dataset. In this work, the CBA classifier has been evaluated using an alternative method for the initial pruning of rules. The pruning is based on a measure that evaluates how the presence of an item in the antecedent affects the confidence of the rule. Several experiments were performed to check the effect of this method of pruning. The results show a significant decrease in the time employed to obtain the classifier, without lost of accuracy nor increment in the number of rules.