Illegal parking risk prediction with consideration of driving behavior

Accurate prediction of street illegal parking risks helps traffic departments impose real-time penalties on illegal parking incidents and optimize police deployment. Considering that parking behavior constitutes the terminal phase of driving activities, the modeling of users' driving patterns is therefore crucial for accurately predicting parking demand. However, existing studies have predominantly focused on the roadway surroundings and spatiotemporal patterns of historical citation records, neglecting the consideration of individual driving behavioral characteristics. To address this limitation, a novel Illegal Parking risk Prediction framework integrated Driving Behaviors (IPP-DB) is proposed. Within this framework, the driving behaviors are quantitatively characterized through graph embedding technique, which extract latent behavioral features from large-scale trajectory datasets. A deep learning-based prediction model is designed to integrate the extracted driving behaviors and related roadway surrounding data. Experiments demonstrate that the proposed model outperforms baseline models. This model contributes to combating illegal parking and promoting sustainable urban transportation systems.