User Behavior Analysis Method Research Based On Trajectory Data

With the rapid development of the Internet industry,the amount of data has become larger and larger.As a consequence,we have entered the era of big data.The huge data set contains a lot of information,which can be discovered through data mining and data handling.Trajectory data,as a component of dig data,mainly consists of latitude and longitude information,time,and other basic information.In this paper,the trajectory data is analyzed and studied based on the electric bicycle trajectory data,in which there are a lot of user behavior features hidden,such as: user life rules,travel habits,frequent paths and so on.Therefore,this paper makes an in-depth study on the track data of e-bike to mine the user's behavior characteristics.The specific contributions are as follows:This paper uses the correlation between trajectory points to improve the traditional stpping point mining algorithm,by which the stay points in the trajectory are extracted to mine the information of the user's residence.Since the trajectory at the stay points have various characteristics of direction change,the Pearson correlation coefficient is used to calculate the correlation between the trajectory data.Firstly,the candidate stay points are extracted according to the correlation coefficient between the trajectory coordinates;then the candidate stay points are filtered by setting the distance and time thresholds to extract the stay point area,and the minimum surrounding rectangle is used to judge and identify the area near the stay point to improve the accuracy of the recognition area.This paper uses kd-tree to improve the cluster by fast search and find of density peaks(KDFDP algorithm)The algorithm is used to mine the information of user's residence,and the results are visualized by Baidu API.Secondly,the algorithm is used to analyze the characteristics of user's riding speed,and is used to the user's schedule speed is modeled by preference matrix.Based on this,the KDFDP user speed clustering model is established.Experiments results verified the effectiveness of the algorithm.In order to mine frequent routes and reduce the lack of frequent paths,the DBSCAN clustering model is established in this paper;which divides the long trajectories into several shorter straight trajectories,and proposes characteristic trajectories for each cluster to represent the movement trend of this class.Experimental results show the effectiveness of the algorithm.