A Parking Utilization Prediction Approach Based on NARNN Model

The accurate prediction of parking occupancy of urban parking spaces is one of the most critical aspects of urban parking resources management and public travel. In order to accurately depict the time-varying characteristics and grasp the future trend of parking utilization, in this paper, we address the problem of prediction of parking utilization. Firstly, based on the characteristics of the time series of parking utilization, a nonlinear auto-regressive neural network (NARNN) model was developed, along with its training method. Secondly, the proposed model was trained via the empirical parking data involved both weekdays and weekends. Thirdly, based on the proposed benchmark, the effectiveness of the proposed method was analyzed. Finally, the applications and future work for the prediction of parking occupancy were summarized.