Selecting Actionable Patterns from Positive and Negative Sequential Patterns

Abstract

Positive and negative sequential patterns (PNSP) play an informative role in various applications. In this paper, a new method is proposed to effectively select the actionable sequential patterns (ASP) from the PNSPs by segmenting and discriminating elements with sequence. First, it is to locally discriminate adjacent elements and incremental elements in the PNSPs. Second, globally segment and discriminate all the elements with sequences. Third, Markov process is further applied to select the ASP by measuring the interestingness of a sequence. The experimental comparisons on synthetic and real-world databases show that the proposed method is very effective to select ASPs.