Assoziationsanalysen im Rahmen des Data Mining

Association analysis is a data mining technique used to discover patterns in a large dataset. The result of this analysis is rules of the form "If A, then B." Different algorithms can be used to generate these rules. They first search for all item sets that meet the specified minimum support and then create association rules based on a user-defined minimum confidence. For practical analyses, the question arises as to which thresholds are sensible to obtain interesting yet not overly numerous rules. Therefore, additional measures of interestingness are used to evaluate the association rules. However, since these do not consider all factors that affect the composition of a transaction, this work develops approaches that allow for a problem-adequate solution.