Research On Hot Path Analysis Based On Trajectory Data Mining

Nowadays, a large number of spatio-temporal trajectory data have been generated. Many mobile devices can record their positions and then send them to the server. As the topic of "big data" is more and more well known, it is becoming an important problem for trajectory data researches that how to process and analyze large trajectory dataset and how to quickly extract significant values from the dataset.The purpose of this study is to use clustering methods and searching meth-ods to find behavior patterns through the movement of moving objects, such as finding hot path. It is to provide decision support information towards traffic management, urban planning and user surveys.Given a pair of locations, a departure time and a travel time limit, our task is to get a hot path highly fitting the real physical world within the specified travel time limit. We first analyze the change of edge hotness in different time ranges by learning historical GPS trajectories, and then propose a hotness function construction method. Next, we propose an effective algorithm with improved strategies to answer the hot path query. Sufficient experiments on a real dataset verify that our methods outperform the comparing algorithms in terms of both effectiveness and efficiency.