A fuzzy approach to prioritize DEA ranked association rules

Abstract

Association rule mining discovers interesting information from large databases. Frequency, reliability and domain knowledge form the multiple criteria for evaluation these association rules. Data envelopment analysis (DEA) is a popular technique used to rank association rules based on the previously mentioned multiple criteria. A decision maker might be interested to have a priority list of these ranked rules based on business and situational requirements. We present an approach to prioritise DEA ranked association rules based on the preference and desirability of the decision maker for different criteria. A modified generalised fuzzy evaluation method (MGFEM) obtains vector-valued fuzzy scores of a group of decision makers and aggregate them to form a preference. A fuzzy logic-based decision support mechanism prioritises these rules based on the decision maker's desirability using the membership function and preference obtained from MGFEM. An example of DEA ranked association rules is presented to explain this innovative approach for prioritisation.