Research On Travel Cycle Activity Pattern Analysis And Location Prediction Supported By User GPS Trajectory Data

There is no doubt that the era of big data is coming.In this case,data becomes an intangible resource,but how to find value from these resources has become a hot spot.Big data contains all kinds of information of users,such as user's GPS trajectory,consumption habits,behavior characteristic data,etc.we can collect important information to carry out big data analysis,further excavate the value of user's diversity,so as to provide more targeted services.With the continuous development of positioning technology,vehicles,mobile phones,smart bracelets and other devices with GPS receiving modules are collecting users' spatiotemporal track data all the time.These user's GPS track data sets show the movement law of a moving object in the nature in the form of data,so there is sufficient space-time information in GPS track data.In addition to providing services for the user's current location,these information can also predict the next location that the user will arrive at,and at the same time,it is more helpful to develop new services and applications.The purpose of this paper is to mine potential user cycle activity patterns from historical user GPS track data,and combine them with similarity to predict the future travel of users.This paper analyzes the advantages and disadvantages of the traditional methods of periodic pattern mining and user location prediction,and puts forward a method of combined prediction of user travel periodic activity pattern mining and travel location based on time matching.First of all,preprocess the user's history track,mine the user's important position points,reconstruct the history track on the user's important position points,transform it into binary sequence,detect the binary sequence cycle as the cycle of the user's important position by Lomb scargle cycle discovery method,and then detect the cycle with the same cycle In this period,each important position will be time matched,high frequency time period will be mined,and periodic activity mode will be formed by connecting them according to time line.Secondly,we calculate the transfer probability between the important location points in the periodic activity mode,and introduce the similarity calculation method to establish a combined location prediction method to predict the important location that the user will arrive.Finally,due to the large distance between the important location points,it is unable to predict the movement of users in detail.In this paper,the roads in the study area are vectorized,and the turning point of the road centerline and the intersection center point are used as the user prediction points between the important location points to build the road network,so as to achieve the detailed prediction between the important location points.In this paper,the real trajectory data set collected daily is used for experiments.The results show that this method can effectively discover the periodic activity pattern of users,and add time matching to get the details of users' movement in each time period;in order to detect the combined prediction method of users' location,this paper uses the test track set to test,and can effectively predict the next location of users' location,which reduces the impact from the transfer probability,It makes the position prediction more accurate and effective.