Title: Prediction and Analysis of Urban Mobility Based on Attention Mechanism and Geographic Information Embedding
Abstract
Predicting and analyzing people's mobility patterns in cities can optimize transportation systems and reduce congestion. Existing studies have ignored the implicit correlation between location and spatiotemporal information, limiting the model performance of location prediction accuracy. This chapter presents a location prediction framework that incorporates attention mechanisms and geographic information embedding method. This framework is validated through cases of location prediction and inference, effectively demonstrating the capability to capture the complexities of urban mobility. By effectively integrating the semantic information of urban spatial data, this framework is able to capture and predict human travel purpose with greater performance. The analysis of the model's predictive performance during specific time periods has revealed an important correlation. It shows that the predictability of human activities is linked to the complexity of the travel. This chapter reviews and analyzes human activity prediction methods, which hold significant theoretical and practical value for fields such as urban planning, traffic management, and public safety.
Full Text Download
Q.E.D.