Research On High Utility Negative Sequential Patterns Mining

Utility-based sequential pattern mining is a new topic in the field of data mining,utility-based sequential pattern mining plays more and more important role in many applications.At present,some utility-based sequence pattern mining algorithms have also been proposed in the industry.However,these algorithms only consider Positive Sequential Pattern(PSP)mining and do not consider Negative Sequential Patterns(NSP).In actual life and applications,those missing items in negative sequential patterns may provide more valuable information.Therefore,the main content of this paper is the research on high utility negative sequential patterns mining.1.Utility-based Negative Sequence Pattern Mining Algorithm.In this paper,we propose a new utility-based negative sequential pattern mining algorithm HUNSPM(High Utility Negative Sequential Patterns Mining)with the e-NSP algorithm and USpan algorithm.First,we use the USpan algorithm to mine all high utility positive sequential patterns.And then,the high utility negative candidates is generated on the basis of the obtained high utility positive sequential patterns,and a new pruning strategy is used to delete high utility negative candidate sequences that do not meet the conditions.Finally,the utility of the negative candidates are calculated by using the negative sequence utility calculation method,and the sequence satisfying the minimum utility threshold is selected.The algorithm mainly solves two key issues: how to generate negative candidates and how to calculate the utility of high utility negative sequences.2.Negative Sequence Pattern Mining Algorithm with Multiple Minimum Utility.The HUNSPM algorithm mines the sequence pattern with a single minimum utility threshold,and it is difficult to set a precise minimum utility.Therefore,this paper also proposes a MUNSPM(Multiple Utility Negative Sequential Patterns Mining)algorithm for mining negative sequence models with multiple minimum utility.This algorithm is based on the HUNSPM algorithm and adds a negative sequential minimum utility threshold setting method,so that we can select the negative sequential patterns based on multiple minimum utility,and provide more comprehensive information for manager.