Research On The Methods Of Selecting Actionable Positive And Negative Sequential Patterns

Positive sequential pattern(PSP)refer to sequences with occurring elements,while negative sequential pattern(NSP)refer to sequences with non-occurring and occurring elements,which can play an irreplaceable role in deeply understanding and tackling many business applications.But some problems occur after mining NSP.Among these problems,how to select actionable positive and negative sequential patterns is the most urgent problem to be solved.Because PSP may be misleading,as in the earlier study of association rules,it is found that positive association rules can be misleading.Although there are some research on actionable knowledge discovery,such as selecting actionable positive and negative association rules and actionable patterns,due to the NSP mining is still in the initial stage,no studies have involved in the selection of actionable positive and negative sequence patterns.In this paper,based on the actionable knowledge discovery and positive and negative association rules,the innovative contents we have done are as follows:1.Analysis the interesting itemsets selection method and improvement method.Analysis of a method for pruning uninteresting itemsets called Wu's method and an improved method SAP based on adjacent elements for selecting actionable positive and negative sequential patterns.2 We propose a new method SAPNSP,SAPNSP improves Wu's method based on sequence increment for selecting actionable positive and negative sequential patterns.3.We propose a method SACPD for selecting actionable positive and negative sequential patterns.For Wu's algorithm can not deal with the sequence,we improve Wu's method by sequence segmentation to fit for sequence,then achieving the purpose of selecting actionable positive and negative sequence patterns.4.We propose a method SAPBN for selecting actionable positive and negative sequential patterns.The specific approach is that through judging the relationship value calculated by the Markov process calculation between the internal elements of a PSP or NSP,then achieving the purpose of selecting actionable positive and negative sequence patterns.