Abstract
Recently, an approach has been proposed in the literatureto extract frequent patterns which occur periodically. Inthis paper, we have proposed an approach to extract rareperiodic-frequent patterns. Normally, the singleminsupbased frequent pattern mining approaches like Apriori andFP-growth suffer from “rare item problem”. That is, at highminsu p, frequent patterns consisting of rare items will bemissed, and at lowminsu p, number of frequent patternsexplode. In the literature, efforts have been made to extractrare frequent patterns under “multiple minimum supportframework”. It was observed that the periodic-frequentpattern mining approach also suffers from the “rare itemproblem”. In this paper, we have extended “multiple mini-mum support framework” to extract rare periodic-frequentpatterns and developed a new algorithm to extract rareperiodic-frequent patterns. Experiment results show thatthe proposed approach is efficient.
