A review of trajectory data preprocessing and mining technology research
Honghao Sheng, Tianqi Wang, Yi Luo, Hongxiang Liang · 2024
Under the background of wireless communication, Internet, and bigdata becoming increasingly perfect, trajectory data of a large number of movable objects are being generated. The trajectory data contains a large amount of spatiotemporal feature information, and by mining it, we can get the information implied in it, such as personnel activity characteristics, interests and hobbies, as well as public transportation movement trajectory and stay characteristics. This paper takes the trajectory data mining processing process as a framework and summarizes the data processing techniques involved in each process. Firstly, it introduces the definition, characteristics and research value of trajectory data; secondly, it describes the preprocessing techniques of trajectory data, including data cleaning, compression, segmentation and matching; thirdly, it summarizes and categorizes the trajectory datamining techniques in the past 5 to 10 years, including support vector machine, artificial neural network, Bayesian network and other methods; finally, its future research directions are prospected.