Research Article
Mobility Prediction Based on POI-Clustered Data
@INPROCEEDINGS{10.1007/978-3-030-00557-3_7, author={Haoyuan Chen and Yali Fan and Jing Jiang and Xiang Chen}, title={Mobility Prediction Based on POI-Clustered Data}, proceedings={Machine Learning and Intelligent Communications. Third International Conference, MLICOM 2018, Hangzhou, China, July 6-8, 2018, Proceedings}, proceedings_a={MLICOM}, year={2018}, month={10}, keywords={Point of interest Clustering Mobility trajectory prediction}, doi={10.1007/978-3-030-00557-3_7} }- Haoyuan Chen
Yali Fan
Jing Jiang
Xiang Chen
Year: 2018
Mobility Prediction Based on POI-Clustered Data
MLICOM
Springer
DOI: 10.1007/978-3-030-00557-3_7
Abstract
Predicting users’ mobility trajectories is significant for service providers, such as recommendation systems for tourist routing, emergency warning, etc. However, the former researchers predict the next location merely by observing the past individual trajectories, which usually performs poor in the accuracy of trace prediction. In this paper, POIs (Points of Interest) information is used to adjust the weight parameters of the predicted results, and the rationality and precision would be improved. The cellular towers are firstly classified into seven types of functional area through POIs. Then the target user’s next possible functional area could be speculated, which acts as a supervision of the ultimate prediction outcome. We use the DP (Dirichlet Process) mixture model to identify similarity between different users and predict users’ locations by leveraging these similar users. As is shown in the results, the methods proposed above are highly adaptive and precise when being utilized to predict users’ mobility trajectories.