Research And Implement Trajectory-based Clustering And Behavior Analysis

Driven by Internet Plus,the scale of the mobile Internet,big data and the Internet of Things(IoT)industries is accelerating.Moreover,Mobile-powered apps and location-based service(LBS)platforms provide users with more efficient and convenient services,while using the apps,users also generate data with hidden value.Among them,the user's movement trajectory is more important data,which is mainly composed of time structure and space features.The data often implies rich information such as the user's behavior habits,travel rules and life rhythm.However,the data is very large.How to efficiently process and utilize these trajectory data including rich information such as geographical location,moving time and speed is a problem.In addition,the trajectory data structure is complicated and difficult to handle.But now,with the rapid development of artificial intelligence,machine learning and deep learning provide a high-quality solution for data analysis,so that valuable potential information can be efficiently extracted from the trajectory data.As an important algorithm in the field of machine learning and data mining,clustering algorithm is often used in the field of data mining.Through this background,this paper will combine deep learning to process trajectory data and cluster analysis of the processed data.In this paper,on the preprocessed trajectory data,preprocessing measures in the machine learning domain are used to remove interference factors in data.And this paper uses the deep learning algorithm to extract the potential structural information in the trajectory data.Specifically,the AutoEncoder neural networks are built to learn the most prominent features in the data distribution,reduce the structural complexity of the data and reduce the dimensionality of the data;In the clustering algorithm experiment,this paper selects the metric formula of the algorithm based on the feature maps,and finally uses the DBSCAN clustering algorithm based on density clustering to analyze the trajectory feature maps.