Statistical pruning of discovered association rules

Nowadays mining association rules in a database is a quite simple task; many algorithms have been developed to discover regularities in data. The analysis and the interpretation of the discovered rules are more difficult or almost impossible, given the huge number of generated rules. In this paper we propose a three step strategy to select only interesting association rules after the mining process. The proposed approach is based on the introduction of statistical tests in order to prune logical implications that are not significant.