An Efficient Algorithm to Mine High Average-Utility Sequential Patterns

Abstract

Mining high-utility sequential patterns (HUSP) has attracted increasing attention in recent years. Current studies of HUSP mining, however, don’t consider the length of sequential patterns, which result in the problem of a huge amount of longer sequential patterns in the mining results. In order to obtain more decision-making information, this paper introduces an average utility measure and proposes HAUSPM method to efficiently mine high average-utility sequential patterns (HAUSP) by considering the length of sequential patterns. Extensive experiments show that the total number of HAUSP mined by HAUSPM is fewer than HUSP under the same minimum utility.