Font Size: a A A

The Research On User Behavior Sequential Patterns Based On Multi-order Markov Chain

Posted on:2019-07-22Degree:MasterType:Thesis
Country:ChinaCandidate:Y ShenFull Text:PDF
GTID:2359330542498210Subject:Management Science and Engineering
Abstract/Summary:PDF Full Text Request
With the development of mobile Internet technology and the rapid popularization of hardware and software,various types of mobile application services are flourishing and have become an indispensable part of people’s daily life.At present,people can complete most activities through mobile devices and do not need to rely on the computer terminal as in the past.Every day a large number of users use the mobile internet business,along with the huge accumulation of user behavior data,which describes in detail the type of business,the time point,and the geographic location when using the business.Big data technology and its application have been gradually popularized within the enterprise,and enterprises also pay more attention to their own data assets to guide their marketing strategy or product design in order to enhance their competition in the market.In this paper,the users can be divided into groups,which can help enterprises to better understand their user types and provide more personalized services.In this paper,a user classification method is proposed,which uses the behavior data of mobile application service to construct the user behavior model.A new multi-order Markov model is introduced to help make up the weakness of the traditional Markov model only considering randomness without only timeliness,thus the model results can better explain the complex behavior of users.By differentiating these user behavior models,the distance matrix between users can be obtained.Then using clustering algorithm to cluster the distance matrix,we can get different types of user groups to filter out the representative users of each group.By analyzing the behavior rules of the representative users,we can use it to characterize the user groups characteristics.Finally,based on these behavioral characteristics,the paper analyzes and summarizes the experimental results,and proposes management suggestions for the enterprise from the business point of view and the user point of view,which is of great guiding significance for enterprises in product and business design in the mobile application market.
Keywords/Search Tags:mobile device business, user behavior analysis, multi-order Markov chain, user clustering
PDF Full Text Request
Related items