Abstract:

Classification and association rule discovery are important data mining tasks. Using association rule discovery to construct classification systems, also known as associative classification, is a promising approach. In this paper, we survey different rule pruning methods used by associative classification techniques. Furthermore, we compare the effect of three pruning methods (database coverage, pessimistic error estimation, lazy pruning) on the number of rules derived from different classification data sets. Results obtained from experimenting on fourteen data sets from the UCI data collection show the need for additional constraints during pruning in associative classification in order to decrease further the size of the resulting classifiers. The results also indicate that lazy pruning algorithms generate very large number of rules if compared with other associative algorithms. This may reduce their use for data mining applications where a concise set of rules is required.