Bikeshare as a Feeder Mode to Metro: Where, When, Who, and Why?
Chenyu Yi, Southeast UniversityShow Abstract
Xinwei Ma, Southeast University
Yanjie Ji, Southeast University
Yang Xu, Ludong University
Yang Liu, Southeast University
The primary objective of this study is to derive a reproducible methodology that isolates bicycle-metro transfer trips using smart card data. Three recognition rules proposed are a maximum transfer time of 10 min, a maximum transfer distance of 300 meters and at least 30 transfer trips occurred at the transfer pair over three consecutive weeks. And then “Return-Enter” and “Exit-Lease” transfer modes are systematically compared on the aspect of the spatial, temporal and demographic circumstances to answer where are the two transfer modes disturbed and how the spatial structure of transfer trips evolves as time goes by; who make the transfer trips and most importantly whether the transfer trips from varied demographic categories have different patterns. The results show that over 89% passengers recognized have less than 6 transfers in 3 weeks. Two transfer modes reveal a distinct commuting characteristic: morning and afternoon peak periods contribute large number of transactions of the whole day, and transfer trips during morning peak period are larger than in the afternoon peak period. Specifically, transfer trips hit the highest at the stations located in urban areas, Hexi district and terminal metro stations that record more transactions. And residential-area-based transactions are more made by female and passengers aged over 35 while office-area-based transactions more by male and nonnative users. The results raise concerns of how to attract more transfer users and improve transfer service. Policy implications are discussed from four aspects: public bike station configurations, dispatch, policy and infrastructure.
A Longitudinal Study of Bike Infrastructure Impact on Bikeshare System Performance
Jia Xu, New York UniversityShow Abstract
Joseph Chow, New York University
The sustainability of bike share systems depends on the bicycle network connectivity and accessibility. However, studies of bike lane infrastructure investment impacts of bike-share ridership have been limited to station-level ridership for supporting resource allocation decisions for bike-share operators. City planning agencies also need to forecast and analyze bike share demand in order to make investment decisions of bike facilities over time. To measure the marginal cost of building bike lanes on bike share demand at a network-wide level over time, an autoregressive conditional heteroscedasticity (ARCH) with autoregressive (AR) disturbance model is proposed to capture system-wide bike ridership. The model is applied to investigate the relationship between the Citi Bike average daily trip counts and the total length of bike lane in New York City. Our results show that installation of one additional mile of bike lanes can lead to an increase of 102Citi Bike daily trips. We demonstrate that this model, as opposed to the previous study in the literature developed at the station level, can provide new insights into system-level causality and temporal lag characteristics. Scenario analysis considering whether a city agency should invest a set of bike lanes immediately or stage them over multiple weeks suggests there is a high effective “discount rate” for ridership benefits from bike lane investments due to temporal dependencies that need to be considered in timing decisions.
Spatial Modeling of Bikeshare Trip Generation in the Twin Cities
Jueyu Wang, University of Minnesota, Twin CitiesShow Abstract
Many U.S. cities have adopted bike share programs because they have potential to increase mobility, diversify transportation mode options, reduce traffic congestion and bring health benefits. Nice Ride Minnesota began operations in 2010 and now supports nearly one-half million trips annually. Using publicly accessible data about bike trips per station per day, this study models correlates of trip generation, with a focus on land use and transportation infrastructure variables. Separate trip generation models are estimated for different types of users. Spatial random effect models are estimated to take into account spatial-autocorrelation between bike stations. Specifically, spatial lag and spatial autoregressive disturbance components are incorporated into the general random effect models. The results reveal heterogeneous impacts of elements of the built environment on trips taken by different types of users. Stations located in the areas with higher population density and job accessibility, a higher percentage of recreational land use, and within a university campus tend to have higher use by members. Stations located in the areas with higher job accessibility, a higher percentage of retail and recreational land use, or close to urban trails have higher use by casual users. The results have implications for siting new stations, planning around stations, and campaigns to increase bike share participation.
How Spatiotemporal Patterns of Biking Behavior Vary Among Different Groups of Bikeshare Users: The Perspectives on Gender, Age, Smart Card Type, and Place of Birth
Xinwei Ma, Southeast UniversityShow Abstract
Mingyuan Yang, Southeast University
Yanjie Ji, Southeast University
Yuchuan Jin, Southeast University
Xu Tan, Southeast University
To understand the travel characteristics and space-time distribution of different groups of public bicycle users, smartcard data were obtained as their travel records in Nanjing, China. This paper takes Data cube, an online analytical processing (OLAP) method for organizing and representing multi-dimensional data, as the main methodology to explore and visualize the space-time travel patterns mined from the bikeshare database. We extend and modify the traditionally three-dimensional data cube into four dimensions, which are Space, Date, Time, and User, develop a user-specified hierarchy for each dimension, and use the number of transactions and the travel time as two quantitative measures. We also generalize the common operations on a multi-dimensional data cube. The visualizations of two-dimensional slices of the data cube mainly show some difference in space-time travel patterns among different groups of bikeshare users. The results suggest that there exists obvious transaction peaks during the morning and afternoon rush hours on weekdays, while the volume at weekends has an approximate uniform distribution. Bad weather condition significantly restricts the bikeshare usage. Besides, seamless smartcard users generally take a longer trip than exclusive smartcard users; and foreign users ride faster than local users. These findings not only support the applicability and efficiency of data cube in the field of visualizing massive smartcard data, but also raise equity concerns among bikeshare users with different demographic backgrounds. Policy and planning implications are involved for Chinese cities to achieve more equal and user-friendly bikeshare systems.
Linking Bikeshare Trips to Light Rail Use in the Minneapolis–St. Paul Region
Eric Barber, University of WashingtonShow Abstract
Rochelle Starrett, University of Washington
Bikeshare systems have frequently been cited as a way to increase transit use by providing transit users with a convenient means to access their final destinations. This study focuses specifically on the Minneapolis-St. Paul, Minnesota, light rail system and the Nice Ride bikeshare system. Data from the 2016 bikeshare season and actual train arrival information from Metro Transit was used to examine how the number of bikeshare checkouts changes in relation to the time since the last train arrival at both a primary group of bikeshare stations, located near light rail stations, and a control group of bikeshare stations. This study fits a negative binomial model for the number of bikes checked out with month, day of week, station, and time since last train arrival as categorical predictor variables and time of day as a continuous predictor variable. Includingtime since last train arrival as a categorical variable improved the model fit over the baseline model and this variable demonstrated a significant and positive effect on the number of bikes checked out, particularly in the first four minutes since the last train arrival. These effects were not observed in the control group, indicating thattime since last train arrival is correlated with the number of bikes checked out. This study highlights the potential of using a data-driven approach to understanding the relation between bikeshare and light rail use, further informing planners and engineers of their synergistic benefits.
Multilevel Urban Form and Bikesharing: Insights from Five Bikeshare Programs Across the United States
Arefeh Nasri, University of Maryland, College ParkShow Abstract
Hannah Younes, University of Maryland, College Park
Lei Zhang, University of Maryland, College Park
Bikesharing programs in their current form have been in place for several years in many cities across the United States. Encouraging people to use bikesharing for their daily routine travels has numerous social, economic, environmental, and health benefits. Therefore, it is important to understand factors influencing bikesharing usage in different urban areas in order to improve the system and encourage more use. This paper investigates how built environment at both local and regional scales influences bikesharing usage in five large metropolitan U.S. areas in the U.S. The data consists of around 9 million bike trips in over 1,500 stations over a one-year period. Regression models are built to predict the number of trips originated from each station with respect to the station’s built environment pattern, as well as the overall urban form in the entire city. The results are consistent with previous research on the effect of land use at the local level on bikesharing demand. At the regional level, results suggest that the overall walkability and job accessibility via bikesharing networks are significant factors influencing bikesharing activities and demand. Models developed in this study could be applied to other communities that are seeking to improve and/or expand their bikesharing systems, as well as cities planning to launch new bikesharing programs.
Quantifying the Effect of Various Features on the Modeling of Bike Counts in a Bikesharing System
Huthaifa Ashqar, Virginia Polytechnic Institute and State UniversityShow Abstract
Mohammed Elhenawy, Virginia Polytechnic Institute and State University
Ahmed Ghanem, Virginia Polytechnic Institute and State University
Mohammed Almannaa, Virginia Polytechnic Institute and State University
Hesham Rakha, Virginia Polytechnic Institute and State University
Bike-sharing systems are an important part of urban mobility in many cities and are sustainable, environmentally-friendly systems. Since the demand of bikes in stations is still not well studied, this paper introduced an effective approach to quantifying the effect of various features on the prediction of bike counts at each station in the San Francisco Bay Area Bike Share. The Random Forest technique was used to rank the predictors, then guided forward step-wise regression and Bayesian Information Criterion were used to develop and compare bike share prediction models, respectively. The final results demonstrate that the time-of-day, temperature, and humidity level (which has not previously been studied) are significant count predictors. It also shows that weather variables are geographic location dependent. Additionally, findings show that variability of bike counts at some stations is not critical if used as a predictor in the regression.
Do New Bikeshare Stations Increase Member Use? A Quasi-Experimental Study
Jueyu Wang, University of Minnesota, Twin CitiesShow Abstract
Greg Lindsey, University of Minnesota, Twin Cities
Over 1000 cities across the world have established bike share programs. Simultaneously a growing number of studies examined correlates of bike share use but few examined the behavior of bike share users. Additionally, research has shown that accessibility is correlated with positively station use. However, most research to date has been cross-sectional analysis and therefore inadequate to establish a causal relationship between accessibility and frequency of use. Using a five-year panel data set of members’ bike share trips from 2010 to 2015 in Twin Cities, we employ a quasi-experimental, difference-in-difference modeling approach to explore the causal relationship between accessibility and frequency of bike share use. Improvements in accessibility are measured as reduction in distance to stations resulting from new stations or relocation of old stations. We find a significant negative impact of distance on frequency of use. Specifically, we find that increasing bike share accessibility has a larger impact in areas with dense bike share services. Moreover, by incorporating built environment variables and interaction terms of built environment with distance, we demonstrate the heterogeneous effects of distance across different built environment contexts. Members who lived in areas with more bike facilities, higher population density, higher percentages of retail, recreational and industrial land uses tend to increase their bike share use more. In terms of bike station densification, installing stations close to members who lived in areas with higher office land uses also increased their bike share use more. The paper concludes with a discussion of implications for policy and planning practice.
Pioneering a State of Good Bikeshare Repair: Defining Asset Management Processes for the District's Capital Bikeshare System
Benito Perez, District Department of TransportationShow Abstract
Carlos Jones, District Department of Human Resources
Matthew Weber, District Department of Energy and Environment
Brahim Sidi M'hamed, District Department of Transportation
The District is enjoying rapid growth in cycling, evident through trends in Census mode share data and presence of more cyclists out on the street. The District Department of Transportation (DDOT) has spent significant resources in the past two decades to improve active transportation planning, outreach, and infrastructure delivery. These efforts have earned the District recognition as a “Cycling Friendly” city. One of those investments is the creation and growth of District’s Capital Bikeshare system. As one of the oldest bikeshare systems in the nation, Capital Bikeshare’s assets are approaching manufacturer specified end of useful life. The District of Columbia’s portion of the system has over 250 stations and 2,700 bicycles in operation today, and owns additional equipment that is awaiting deployment. The lifecycle of assets is dependent on utilization and a variety of other variables. For bikeshare asset management, there is limited precedent on how to best assess the State of Good Repair and determine budget forecasts for capital replacement. This paper intends to define a bikeshare State of Good Repair for the Capital Bikeshare system, evaluate historical bike and station maintenance records and utilization data, collect and assess aesthetic condition of the assets, understand state of repair trends and patterns for state of good repair forecasting, and determine a business approach to sustainably assess the condition of the District’s Capital Bikeshare equipment. This exercise will also inform future asset management planning and investment for the system.
Static Repositioning in Bikesharing Systems with Broken Bikes
Yue Wang, University of Hong KongShow Abstract
Wai Yuen Szeto, University of Hong Kong
The Bike Sharing System (BSS) has been receiving increasing popularity in transportation plans. To distribute bikes reasonably across stations to cater for the time-related demand, bikes need to be moved among stations using a fleet of vehicles. This process is the bikes repositioning. Besides the transshipment of good bikes, all broken bikes need to be carried back to the depot to repair. This research aims to handle both good and broken bikes in a bike sharing network in order to achieve the perfect balance between demand and supply and make sure all broken bikes are moved back to the depot. The objective of this repositioning operation is to minimize the total CO 2 emission of the repositioning vehicle during the whole journey. In this research, a Mixed Integer Linear Program (MILP) model is presented to describe the problem mentioned above and a commercial solver is used to solve it. Problem characteristics are dissected by analyzing factors that would affect emissions. The results indicate that the proposed model can describe the problem properly and the emission will be affected according to different constraint settings.
A Planning Assessment of Bikesharing Systems: Top-Down or Bottom-Up Demand Practices?
Alec Biehl, Northwestern UniversityShow Abstract
Alireza Ermagun, Northwestern University
Amanda Stathopoulos, Northwestern University
This research juxtaposes station-level and community-level approaches to model and estimate the Annual Average Daily Bicyclist (AADB). We model bikeshare demand from 459 Divvy stations in Chicago between June 1, 2015 and May 31, 2016 at two spatial scales. Elasticity calculations and prediction error comparisons reveal the importance of both built-environment and socio-demographic variables in bikeshare modeling and the development of context-sensitive interventions. The detailed comparison of different levels of aggregation for analysis of bikeshare demand and user impact highlights that each level contributes insights to planners and policymakers. While disaggregate data provides the most information for planners in terms of improving bikeshare systems, there is value in adopting an aggregated approach for transport policy that accounts for potential neighborhood effects. In addition, the control for socio-demographic factors around stations pinpoints the variation in socio-spatial effects that planners need to account for when measuring outcomes and equity impacts.
The Impact of Weather Condition and Built Environment on Public Bikesharing Trips in Beijing
Jiancheng WengShow Abstract
Pengfei Lin, Beijing University of Technology
Quan Liang, Beijing University of Technology
Dimitrios Alivanistos, Vrije Universiteit Amsterdam
Siyong Ma, Beijing University of Technology
As bicycling regains popularity around the world, the Beijing Public Bikesharing System, launched in 2012, allows users to access shared bicycles for short trips. After five years of operation, while the system is widely used, it faces the problems of bike unavailability and dock shortage at various stations due to the tidal characteristics of bicycle travel. It’s necessary to investigate the influence of different weather conditions and nearby built environment of stations on bikesharing trips. Using historical trip data from 2016, concerning 543 stations in Beijing, log-linear regression models are developed to estimate the impact of daily weather and time events on trip generation and attraction. Moreover, the effects of built environment variables such as land use and transport infrastructure are investigated both on workday and non-workday ridership at the station-level. The results indicate temperature is not linearly associated with daily ridership. Daily usage decreases according to rainfall, snowfall, wind speed and weekends/holidays. Light and heavy pollution have no significant influence on bikesharing demand; however, severe pollution has a negative influence on usage. The effect of transport infrastructure (subway stations, bus stops and bikeway length) is crucial in increasing bikesharing demand. The number of residence locations and shopping locations is generally associated with usage. Proximity to colleges does not show an obvious ridership increase, which is different from other cities. Parks encourage bikesharing usage during weekends/holidays time than workdays. The findings may help planners or managers to design and modify public bikesharing stations effectively, increasing usage while reducing rebalance costs.
Effectiveness of Small-Scale Bikesharing Systems According to the Analysis of Turnover Station Ratios
Pilar Jiménez, Universidad Politécnica de CartagenaShow Abstract
Maria Nogal, Trinity College, Dublin
Brian Caulfield, Trinity College, Dublin
Many studies about Bike Sharing Systems (BSS) show the good performance of these schemes when bicycles are highly used on a daily basis. Most of them analyse large cities such as Washington D.C., Melbourne, London, Barcelona or Dublin, which have more than 500,000 inhabitants. Nevertheless, studies on BSS in small cities with a population between 20,000 and 100,000 inhabitants are still undeveloped. This paper is focused on BSSs in small cities with the aim of shedding light on how to improve BSS performance in small cities based on current experiences. More precisely, the paper analyses how the BSSs of two Irish cities, Limerick and Galway, have worked during the first two years of implementation (2015-2016). It is focused on the effectiveness of the stations using the Turnover Station ratio, which measures the usage degree of each station. This analysis will allow the identification of the stations working correctly, which stations need a deep study to improve its resources (bikes and docks) and if some station should be removed. Moreover, these findings will result in increasing knowledge about the implementation and development of small scale BSSs. For example, some of the main results point out that to assess the successful of these schemes in small cities the daily basis for analysing the usage should be changed on the monthly basis in order to consider an overview of the systems and, the differences between annual trips and temporary trips also affect the classification of stations according to the turnover ratio.
Predicting Bike Availability in Bikesharing Systems Using Dynamic Linear Models
Mohammed Almannaa, Virginia Polytechnic Institute and State UniversityShow Abstract
Mohammed Elhenawy, Virginia Polytechnic Institute and State University
Hesham Rakha, Virginia Polytechnic Institute and State University
The significant increase in the use of bike sharing systems (BSSs) causes imbalances in the distribution of bikes, creating logistical challenges. Moreover, imbalanced BSSs discourage bike riders who find it difficult to pick up or drop off a bike at their desired location. The first step to solve this logistical problem is forecasting the bike availability at each station in a Bike Sharing System. Forecasting the bike count is a challenging task because the developed models have to consider the change in the demand patterns at each station in the BSS, caused by the interaction between the incentivized users and the BSS. This paper adopted DLMs to model and predict bike counts in a Bike Sharing System. We proposed using first and second order polynomial models because of their simplicity. Both DLMs are applied to a Bike Sharing System of 70 stations in the San Francisco Area for the year of 2014-2015. Different prediction windows were used: 15, 30, 45, 60 and 120 minutes. The adopted models are shown to be powerful modeling tools in capturing temporal evolution in the underlying demand pattern and hence the bike count at BSS stations. The obtained results show the first and second DLMs are able to predict the bike count with a prediction error of 0.37 bikes/station for a 15-minute prediction horizon (corresponding to a percentage error of 5%) and 1.1 bikes/station for a 2-hour prediction horizon. Comparison results show that the DLMs outperform the Least-Squares Boosting algorithm. Besides it is comparable to the Random Forest model predictions for 15- and 30-minute prediction horizons.
Forecasting the Travel Demand of the Station-Free Sharing Bike Using a Deep Learning Approach
Chengcheng Xu, Southeast UniversityShow Abstract
Junyi Ji, Southeast University
Pan Liu, Southeast University
The station-free sharing bike is a new sharing traffic mode that has been deployed in a large scale in China in the early 2017. Without docking stations, this system allows the sharing bike to be parked in any proper places. This study aimed to develop a dynamic demand forecasting model for station-free bike sharing using the deep learning approach. The spatial and temporal analyses were first conducted to investigate the mobility pattern of the station-free bike sharing. The result indicates the imbalanced spatial and temporal demand of bike sharing trips. The long short-term memory neural networks were developed to predict the bike sharing trip production and attraction at TAZ for different time intervals, including the 10-min, 15-min, 20-min and 30-min intervals. The validation results suggested that the developed LSTM NN models have reasonable good prediction accuracy in trip productions and attractions for different time intervals. The ARIMA models were also developed to benchmark the LSTM NN. The comparison results suggested that the LSTM NN models provide better prediction accuracy than the ARIMA model for different time intervals. The developed LSTM NN can be used to predict the gap between sharing bike inflow and outflow, which provide useful information for rebalancing the sharing bike in the system.
Characterizing Temporal Variations of Public Bike Riding: A Case Study of Ningbo, China
Chenxi Lu, Ningbo University of TechnologyShow Abstract
Jing Bie, University of Nottingham
This paper analyzes the temporal variations of public bike riding over the hours, the days and the months. A whole year of usage data (34,098,787 trips) are collected for the bikesharing system in Ningbo, China. Data analysis shows significant seasonal variations of bike rental amounts. The seasonal reductions of bike riders are explained mainly by three factors: nonpermanent residents leaving the city during the Chinese Spring Festival, uncomfortable high/low outdoor temperature, and rainy/snowy weather. Weather impact analysis suggests that rainfall decreases rental amounts significantly, with 1mm of daily rainfall expected to reduce daily rental amount by about 10%. The results also show significant day-of-the-week variations, with rental amounts in weekends 20-30% lower than in weekdays, indicating a clear commuter pattern of usage. The commuter pattern is further proven by the time of day variation with the double rental peaks of morning and afternoon. It should be noted that the total usage in either morning or afternoon rush hours only accounts for 60% of the total bike supply, which indicates that some bikes in the system may have been positioned in the wrong place at peak times. It is further shown that peak hour usage accounts for only 10~12% of AADT, much lower than the peak hour percentages for other modes. These temporal characteristics imply that the bike turnover rate could increase further if the efficiency of the bike supply is improved with proper location and relocation of bikes.
A Spatial-Temporal Analysis of Baltimore Bikeshare Ridership
Amirreza Nickkar, Morgan State UniversityShow Abstract
Celeste Chavis, Morgan State University
Handy Phyall, Morgan State University
Philip Barnes, University of Delaware
Bike share programs are growing in popularity across the United States as cities aim to provide an alternative mode of transportation that addresses last mile needs while promoting recreation and tourism. In fall 2016, Baltimore launched its bike share with 20 stations. Using trip data for over 17,000 trips from October to May, this study explores the temporal and spatial patterns of bike share use in the city. The study found that there were distinctly different patterns in bike share use on weekend and weekdays. Weekend trips were dominated by trips starting and ending in recreational and commercial land uses whereas weekday trips largely occurred at administrative and job centers. Additionally the study determined that non-members have longer trip duration and take more touring trips – trips which start and end at the same station. The study suggests there are two distinct user groups of Baltimore’s Bike Share: (1) subscribers who use this service for commuting and last mile purposes and (2) occasional users that utilize bike share for recreation.
A GIS Performance Measure-Based Methodology for Prioritizing Bikeshare System Expansion
Nicholas Jehn, Auburn UniversityShow Abstract
Md Atiquzzaman, Auburn University
Jeffrey LaMondia, Auburn University
Rod Turochy, Auburn University
Cycling is becoming an increasingly important mode of transportation in urban areas. Bike share programs are an integral part of this trend in many communities from large metropolitan areas to college towns. Following a review of lessons learned from feasibility studies and system expansions of several bike share programs, a GIS-based method to identify and prioritize locations for system expansion is developed. The method incorporates factors that describe transportation demand and existing infrastructure as well as demographic and land use factors to identify potential station locations and develop an activity-based accessibility model. This method is then applied using a case study for expansion of the existing War Eagle Bike Share program in Auburn, Alabama. Data collection and processing, calculations of suitability scores for potential station locations, and an approach to prioritizing installation of recommend locations is demonstrated.