Incremental discovery of probabilistic rules from clinical databases based on rough set theory

Abstract

Extending concepts of rule induction methods based on rough set theory, we introduce a new approach to knowledge acquistion, which induces probabilistic rules incrementally, called PRIMEROSE-INC (Probabilistic Rule Induction Method based on Rough Sets for Incremental Learning Methods). This method first uses coverage rather than accuracy, to search for the candidates of rules, and secondly uses accuracy to select from the candidates. This system was evaluated on clinical databases on headache and meningitis. The results show that PRIMEROSE-INC induces the same rules as those induced by PRIMEROSE, which extracts rules from all the datasets, but that the former method requires much computational resources than the latter approach.