Short-term Bus Passenger Flow Forecast Based On Deep Learning

The public transportation system is an essential part of the life of the citizens and it’s the basis of intelligent transportation system(ITS). This paper tries to predict shortterm bus passenger flow by using deep learning approach that called SAE model and DBN model. The model training and evaluation were carried out using the credit card records of the Suzhou bus IC card. The experimental results show that the SAE and DBN models can reduce the prediction error by 9.51% and 10.48%, respectively, compared with the traditional method. The methods of deep learning show a good application prospect in the short-term bus passenger flow forecasting.