Predicting station-level hourly demand in a large-scale bike-sharing network: A graph convolutional neural network approach. (December 2018)

• Record Type: Journal Article
• Authors: Lin, Lei; He, Zhengbing; Peeta, Srinivas
• Is Part Of: Transportation research. Volume 97(2018)
• Page Start: 258
• Page End: 276
• Publication Date: 2018-12
• Subjects: Bike sharing -- Graph Convolution Neural Network -- Data-driven graph filter -- Long Short-term Memory network -- Deep learning -- Spatio-temporal demand prediction; Transportation -- Periodicals; Transportation -- Technological innovations -- Periodicals; 388.011
• DOI: 10.1016/j.trc.2018.10.011
• Languages: English
• ISSNs: 0968-090X

Abstract

Highlights: Propose Graph Convolutional Neural Network with Data Driven Graph Filter (GCNN-DDGF). Propose Recurrent GCNN-DDGF to capture the temporal dependencies. Explore two GCNN-DDGFs that outperforms four normal GCNNs and seven benchmark models. Obtain insights on the "black box" of GCNN-DDGF by conducting graph network analysis. The DDGF uncovers hidden correlations between stations to improve demand predictions. Abstract: This study proposes a novel Graph Convolutional Neural Network with Data-driven Graph Filter (GCNN-DDGF) model that can learn hidden heterogeneous pairwise correlations between stations to predict station-level hourly demand in a large-scale bike-sharing network. Two architectures of the GCNN-DDGF model are explored; GCNNreg -DDGF is a regular GCNN-DDGF model which contains the convolution and feedforward blocks, and GCNNrec -DDGF additionally contains a recurrent block from the Long Short-term Memory neural network architecture to capture temporal dependencies in the bike-sharing demand series. Furthermore, four types of GCNN models are proposed whose adjacency matrices are based on various bike-sharing system data, including Spatial Distance matrix (SD), Demand matrix (DE), Average Trip Duration matrix (ATD), and Demand Correlation matrix (DC). These six types of GCNN models and seven other benchmark models are built and compared on a Citi Bike dataset from New York City which includes 272 stations and over 28 million transactions from 2013 to 2016. Results show that the GCNNrec -DDGF performs the best in terms of the Root Mean Square Error, the Mean Absolute Error and the coefficient of determination ( R 2 ), followed by the GCNNreg -DDGF. They outperform the other models. Through a more detailed graph network analysis based on the learned DDGF, insights are obtained on the "black box" of the GCNN-DDGF model. It is found to capture some information similar to details embedded in the SD, DE and DC matrices. More importantly, it also uncovers hidden heterogeneous pairwise correlations between stations that are not revealed by any of those matrices.