Similarity Analysis Of Users’ Trajectories Based On GPS Data

Nowadays, the networks are having higher demand on the mobilization on the tre nd of global urbanization. With the rapid development and popularity of GPS, people can locate themselves in real time, as well as the trajectories of their movement. There fore, we could get the users’ behavioral information by analyzing their GPS trajectory data, so as to predict their destinations, ways of travelling, and even roads about to suf fer from traffic jams, and that can help alleviate traffic jams for better mobilization. T he work described in this paper is to figure out the information among the users throu gh mining the raw data of GPS, that is, people are able to find others sharing similar movements, which could help to establish their social networks, and predict a user’s d estination through the other’s as well. This can benefit the users by offering some reco mmendations around the destinations like shopping malls, restaurants, amusement par ks, etc.The data mining based on the GPS data has became a hot research in data mining in recent years. The complex of raw GPS data may reduce computational efficiency, so we can dig good features in the raw GPS data, use the good features construct GPS data space. In this paper, we use two methods to construct the feature space, the one is that use the road segment as feature and the another one is that use the road junction as feature. The first method is to use mobile speed and direction as characteristic value, the second method is to use intersection as characteristic value.Extract good features from the raw GPS data to construction the GPS data space. Use the good features can speed up the research and calculate of the users’ trajectory. This article mainly analysis the relationship between the multi-user. In other words, this article mainly analysis the similarity between multiple users. This paper cluster the historical trajectories, put the similar trajectory in a type of cluster. We use mutual information method to cluster the trajectories because the length of users’ trajectories is different.