An effective algorithm for mining interesting quantitative association rules
Keith C. C. Chan, Wai-Ho Au · 1997
In this paper, we describe a novel technique, called APACS2, for mining interesting quantitative association rules from very large databases. To effectively mine such rules, APACS2 employs adjusted difference analysis. The use of this technique has the advantage that it does not require any user-supplied thresholds which are often hard to determine. Furthermore, APACS2 also has the advantage that it allows users to discover both positive and negative association rules. A positive association rule tells us that a record having certain attribute value will also have another attribute value whereas a negative association rule tells us that a record having certain attribute value will not have another attribute value. The fact that APACS2 is able to mine both positive and negative association rules and that it uses an objective yet meaningful measure to determine the interestingness of a rule makes it very effective at different data mining tasks.