Public subsidy of transit services has increased dramatically in recent years, with little effect on overall ridership. Quite obviously, a clear understanding of the factors influencing transit ridership is central to decisions on investments in and the pricing and deployment of transit services. Yet the literature about the causes of transit use is quite spotty; most previous aggregate analyses of transit ridership have examined just one or a few systems, have not included many of the external, control variables thought to influence transit use, and have not addressed the simultaneous relationship between transit service supply and consumption. This study addresses each of these shortcomings by (1) conducting a cross-sectional analysis of transit use in 265 US urbanized areas, (2) testing dozens of variables measuring regional geography, metropolitan economy, population characteristics, auto/highway system characteristics, and transit system characteristics, and (3) constructing two-stage simultaneous equation regression models to account for simultaneity between transit service supply and consumption. We find that most of the variation in transit ridership among urbanized areas – in both absolute and relative terms – can be explained by factors outside of the control of public transit systems: (1) regional geography (specifically, area of urbanization, population, population density, and regional location in the US), (2) metropolitan economy (specifically, personal/household income), (3) population characteristics (specifically, the percent college students, recent immigrants, and Democratic voters in the population), and (4) auto/highway system characteristics (specifically, the percent carless households and nontransit/non-SOV trips, including commuting via carpools, walking, biking, etc.). While these external factors clearly go a long way toward determining the overall level of transit use in an urbanized area, we find that transit policies do make a significant difference. The observed range in both fares and service frequency in our sample could account for at least a doubling (or halving) of transit use in a given urbanized area. Controlling for the fact that public transit use is strongly correlated with urbanized area size, about 26% of the observed variance in per capita transit patronage across US urbanized areas is explained in the models presented here by service frequency and fare levels. The observed influence of these two factors is consistent with both the literature and intuition: frequent service draws passengers, and high fares drive them away.

Nature and/or nurture? Analyzing the determinants of transit ridership across US urbanized areas

Real-time crash risk prediction on arterials based on LSTM-CNN.

Urban flow prediction from spatiotemporal data using machine learning: A survey

The accurate short-term passenger flow prediction is of great significance for real-time public transit management, timely emergency response as well as systematical medium and long-term planning. In this paper, we propose an end-to-end deep learning framework that can simultaneously make multi-step predictions for all stations in a large scale metro system. A sequence to sequence model embedded with the attention mechanism forms the backbone of this framework. The sequence to sequence model consists of an encoder network and a decoder network, making it good at modeling sequential data with varying lengths and the attention mechanism further enhances its ability to capture long-range dependencies. We use the proposed framework to predict the number of passengers alighting at each station in the near future, given the number of passengers boarding at each station in the last few short-term periods. The large quantities of real-world data collected from Singapore’s metro system are used to validate the proposed model. In addition, a set of comparisons made among our model and other classical approaches evidently indicates that the proposed model is more scalable and robust than other baselines in making multi-step and network-wide predictions for short-term passenger flow.

Sequence to sequence learning with attention mechanism for short-term passenger flow prediction in large-scale metro system

A survey on long short-term memory networks for time series prediction

A geographically and temporally weighted regression model to explore the spatiotemporal influence of built environment on transit ridership

Accurate metro ridership prediction can guide passengers in efficiently selecting their departure time and transferring from station to station. An increasing number of deep learning algorithms are being utilized to forecast metro ridership due to the development of computational intelligence. However, limited efforts have been exerted to consider spatiotemporal features, which are important in forecasting ridership through deep learning methods, in large-scale metro networks. To fill this gap, this paper proposes a parallel architecture comprising convolutional neural network (CNN) and bi-directional long short-term memory network (BLSTM) to extract spatial and temporal features, respectively. Metro ridership data are transformed into ridership images and time series. Spatial features can be learned from ridership image data by using CNN, which demonstrates favorable performance in video detection. Time series data are input into the BLSTM which considers the historical and future impacts of ridership in temporal feature extraction. The two networks are concatenated in parallel and prevented from interfering with each other. Joint spatiotemporal features are fed into a fully connected network for metro ridership prediction. The Beijing metro network is used to demonstrate the efficiency of the proposed algorithm. The proposed model outperforms traditional statistical models, deep learning architectures, and sequential structures, and is suitable for ridership prediction in large-scale metro networks. Metro authorities can thus effectively allocate limited resources to overcrowded areas for service improvement.

Parallel Architecture of Convolutional Bi-Directional LSTM Neural Networks for Network-Wide Metro Ridership Prediction

The invention provides an urban public transport passenger flow forecast method and equipment based on in-depth learning, used for solving the problem that an accurate method capable of forecasting future short-term passenger flow on an urban public transport line does not exist. The method disclosed by the invention comprises the following steps: establishing a grid map according to a traffic line and geographic information of stations; updating pixel values of grids corresponding to stations in the grid map according to the passenger flow information of the stations; and by taking the grid map as input of an in-depth learning algorithm, forecasting the passenger flow of the urban public transport. According to the technical scheme, the method disclosed by the invention has the advantagesthat the metro passenger flow station network is converted into picture information to serve as input of a convolutional neural network by utilizing the geographic information system, and the space information of the passenger flow is extracted. Then, the output of the convolutional neural network serves as input of a long/short-term neural network so as to extract time characteristics. Finally,spatial-temporal characteristics serve as input of a full-join neural network so as to perform passenger flow forecast.

Urban public transport passenger flow forecast method and equipment based on in-depth learning

The invention discloses a method for identifying and predicting a bus passenger flow influence factor based on geographically and temporally weighted regression (GTWR). The method comprises steps of: extracting a traffic zone hour bus passenger flow and calculating built environment density; 2, constructing a space-time three-dimensional coordinate system according to the time of a passenger flow observation point and latitude and longitude to calculate space-time distance, and reckoning a spatial regression weight matrix according to a Gaussian function and the distance; 3, calculating a relation between a passenger flow volume and land utilization under different space-time conditions based on the GTWR; and 4, obtaining a change of a relevant parameter to a coefficient according to the calculation to perform visualization processing in time and space, and analyzing an inherent law. The method takes account of the influence of a time factor on the bus passenger flow and the built environment relation, can deeply excavate an internal relation between the passenger flow and the land utilization, accurately predicts the bus passenger flow, and provides the scientific theoretical guidance for the bus line planning and operation management.

Jiyu Zhang

Papers

A geographically and temporally weighted regression model to explore the spatiotemporal influence of built environment on transit ridership

Parallel Architecture of Convolutional Bi-Directional LSTM Neural Networks for Network-Wide Metro Ridership Prediction

Urban public transport passenger flow forecast method and equipment based on in-depth learning

Method for identifying and predicting bus passenger flow influence factor based on geographically and temporally weighted regression

Forecasting subway demand in large-scale networks: a deep learning approach