Association Rule Mining I

Abstract

This chapter looks at the problem of finding any rules of interest that can be derived from a given dataset, not just classification rules as before. This is known as Association Rule Mining or Generalised Rule Induction. A number of measures of rule interestingness are defined and criteria for choosing between measures are discussed. An algorithm for finding the best N$N$rules that can be generated from a dataset using the J$J$-measure of the information content of a rule and a ‘beam search’ strategy is described.