Predicting Passenger's Public Transportation Travel Route Using Smart Card Data

Chen Yang, Wei Chen, Bolong Zheng, Tieke He, Kai Zheng, Han Su

Abstract: Transit prediction is a important task for public transport institutions and urban planners to provide better transit scheduling and urban planning. In recent years, there are a lot of research on traffic prediction, but the existing works focus predicting the monolithic traffic trend, and few works focus on passenger’s public transportation travel route. In this paper, we study the passenger’s travel route and duration prediction. We propose a prediction model based on LSTM neural network to predict passenger’s travel route and duration. Specifically, we leverage multimodal embedding to extract passenger’s features which are highly related to passenger’s travel route and then use a LSTM-based model to improve the prediction accuracy. To verify the effectiveness of our model, we conduct extensive experiments using a real dataset which is collected from Brisbane in Australia for four months. The experimental results show that the accuracy of our model is better than baseline models.