Spatiotemporal Prediction of Urban Online Car-Hailing Travel Demand Based on Transformer Network

Abstract

Online car-hailing has brought convenience to daily travel, whose accurate prediction benefits drivers and helps managers to grasp the characteristics of urban travel, so as to facilitate decisions. Spatiotemporal prediction in the transportation field has usually been based on a recurrent neural network (RNN), which has problems such as lengthy computation and backpropagation. This paper describes a model based on a Transformer, which has shown success in computer vision. The study area is divided into grids, and the structure of travel data is converted into video frames by time period, based on predicted spatiotemporal travel demand. The predictions of the model are closest to the real data in terms of spatial distribution and travel demand when the data are divided into 10 min intervals, and the travel demand in the first two hours is used to predict demand in the next hour. We experimentally compare the proposed model with the three most commonly used spatiotemporal prediction models, and the results show that our model has the best accuracy and training speed.

Exploring the Impacts of COVID-19 and Lockdown on Online Car-Hailing Travel in Shanghai

“…Bi et al [33] successfully identified job-housing sites using online car-hailing data. Notably, machine learning (especially deep learning) methods were also introduced into the study of online car-hailing data, to predict variations in patterns [34], changes in the supply and demand [35], and so on.…”

Short-Term Demand Forecasting of Urban Online Car-Hailing Based on the K-Nearest Neighbor Model

“…In order to predict the travel demand of online car-hailing more accurately, Bi et al (2022) proposed a new spatio-temporal prediction method model based on Transformer architecture. Compared with the three most commonly used models, the results show that the model has the best prediction accuracy and prediction accuracy training speed [ 14 ]. With the rapid development of online big data technology and the rise of non-parametric regression models in the field of forecasting, forecasting methods represented by the K-nearest neighbor method have started to be introduced into the field of transportation.…”