2021

Mohammed Almannaa (almannaa@vt.edu), King Saud University
Faisal Alsahhaf, King Saud University
Huthaifa Ashqar, Booz Allen Hamilton, Inc.
Mohammed Elhenawy, Queensland University of Technology
Mahmoud Masoud, Queensland University of Technology
Andry Rakotonirainy, Queensland University of Technology

E-scooter sharing systems have been emerging in many countries as a new mode of microbility with the goal of solving urban issues such as the first-/last-mile transportation problem. The study explores the feasibility of launching an e-scooter sharing system as a new micromobility mode, mobility-as-a-service, and part of the public transportation system in Saudi Arabia. Therefore, an online survey was conducted in April 2020 to shed light on the perception of e-scooter systems in Saudi Arabia. A representative random sample of 439 respondents was collected (33% female and 67% male) where majority (approximately 73%) indicated willingness to use the e-scooter sharing system if available (males are twice as likely to agree than females). Roughly 75% of the respondents indicated that open entertainment areas and shopping malls are ideal places for e-scooter sharing systems in Saudi Arabia. Results indicated that people who use ride-hailing services, such as Uber and Careem, expressed more willingness to use e-scooters for various purposes. The study found that the major obstacle for deploying e-scooters in Saudi Arabia is the lack of sufficient infrastructure (70%) followed by weather (63%) and safety (49%). Moreover, the study found that approximately half of the respondents believed that COVID-19 will not affect their willingness to ride e-scooters. Results will enable policymakers and operating agencies to evaluate the feasibility of deploying e-scooters and better manage the operation of the system as an integral and reliable part of public transportation in Saudi Arabia.

Yi Jing, Tongji University
Songhua Hu, University of Maryland
Hangfei Lin, Tongji University

Shared dockless scooters are expanding across the world rapidly, serving as a strong alternative to the bikesharing system. Limited studies shed light on the usage characteristics of this emerging service, and relatively little is known about its similarities and differences with bikesharing. This study compares the key factors related to bikesharing and shared scooters as well as their relationships in the city of Minneapolis. A set of negative binomial regressions are built under the framework of structural equation modeling to specify causal mechanisms between bikesharing and scooter sharing across different periods, controlling for exogenous variables including social demographics, transportation facilities, land use, and spatial features. Findings confirm a significantly positive direct effect from bikesharing to scooter sharing, while the reverse effect is insignificant. Besides, bikesharing and scooter sharing are both more popular in areas with higher population density, higher income, more youngsters, and more commercial/public/mixed-use lands. Compared with scooter sharing, bikesharing users ride more for commuting, rely more on bicycle lanes, and have a more stable complementary relationship with transit. Findings are essential to help different shared micromobility services find their corresponding niches, as well as to guide the sustainable development of the mixed shared micromobility system.

Tatsuya Fukushige (tfukushige@ucdavis.edu), University of California, Davis
Dillon Fitch, Institute of Transportation Studies (ITS)
Susan Handy, University of California, Davis

The recent emergence of dock-less electric bike- and scooter-share services have allowed a growing number of cities to look toward bike/scooter share to improve environmental, social, and health outcomes from the transportation system. Increasing bike/scooter share use is likely to increase users’ physical activity and reduce their vehicle miles of travel (VMT) and related greenhouse emissions – if it substitutes for car use. If the major mode shift comes from public transit, owned bike, or walking, the benefits will be more limited. The goal of this paper is to identify the factors, including trip attributes, individual characteristics, and land use characteristics, that influence mode substitution, defined here as the mode that is replaced when bike/scooter share is used. Data come from a two-wave longitudinal survey of bike-share users in the Sacramento, CA region in 2018 and 2019. The analysis shows that long trips and trips that start at non-commercial locations are the most likely to represent car substitution and that some groups, such as women and those who have a private car, are more likely to report car substitution. These results provide guidance for bike-share operations toward the goal of enhancing car substitution.

Nishatee Binte Shahid, Bangladesh University of Engineering and Technology
Sk. Nahia Ahsan, Bangladesh University of Engineering and Technology
Md. Istiak Jahan, Bangladesh University of Engineering and Technology
Md Asif Raihan, Bangladesh University of Engineering and Technology
Sumaiya Afrose Suma, Bangladesh University of Engineering and Technology
Md. Hadiuzzaman (mhadiuzzaman@ce.buet.ac.bd), Bangladesh University of Engineering and Technology

Ride sharing service is comparatively a new concept in Bangladesh, and thus, its safety is yet to be assessed.  Structural equation modeling (SEM) was used to develop an empirical model to identify the relationships between major attributes that affect ride sharing bike service safety. The model was calibrated using a collected data of 451 ride sharing bike users from Dhaka, Chittagong and Sylhet who were interviewed with a formal questionnaire to find out about their experience, satisfaction level and opinion about the existing service as well as their expectations. Depending on their stated preferences, the effect of twenty-three observed variables; and three latent variables, “Perceived Safety”, “Service Quality” and “Crash and Security Potential” were analysed. “Efficiency in Arriving” from the observed variables and “Service Quality’ from the latent variables were found having the highest influence on the ride sharing bike safety. This research would provide the ride sharing bike services with constructive reviews in order to enhance their level of safety to international standards and make the service more acceptable to the users. A clear perception for the implementation of effective government decisions will also then be obtained.

Hongtai Yang (yanghongtai.top1@gmail.com), Southwest Jiaotong University
Yongbo Zou, Southwest Jiaotong University
Yaohu Xiong, Southwest Jiaotong University
Xiaobo Liu, Southwest Jiaotong University

Dockless bike sharing system, as an emerging travel mode, has provided an efficient "last mile" connection for subway systems and expanded their service coverage. In previous studies, distance-decay curve is commonly used to quantify the subway station coverage, for pedestrian mode, which motivates us to fill the scant research on the newly emerged shared bicycles. A trip-chain based method is proposed to identify and extract the set of bicycle trips connecting to subway stations. Thus, consecutive trip pairs for each traveler where the former trip ends in the vicinity of one subway station while the later starts in that of another could be identified. Based on statistic tests, the exponential curve was chosen as the representative distance-decay curve. All curves rise with distance as the beginning and then drops, which indicates walking is still popular for short distance trips. We further explore factors influencing the size of the coverage area based on the 75th percentile of the distance-decay curve. Out of potential influencing factors such as the number of station entrances, straightness, average distance of four nearest stations, morning and evening trip ratio, and transfer station, regression analysis shows that average distance of the four nearest station is the most important variable. This study provides an effective reference to researchers and planners to design and optimize the built environment around subway stations.

Suzanne Maas (suzanne.maas@um.edu.mt), University of Malta
Paraskevas Nikolaou, University of Cyprus
Maria Attard, University of Malta
Loukas Dimitriou, University of Cyprus

Bicycle sharing systems (BSS) have been implemented in cities worldwide in an attempt to promote cycling. Despite exhibiting characteristics considered as barriers to cycling, such as hot summers, hilliness and car-oriented infrastructure, Southern European island cities and tourist destinations Limassol (Cyprus), Las Palmas de Gran Canaria (Canary Islands, Spain) and the Valletta conurbation (Malta), are all experiencing the implementation of BSS and policies to promote cycling. In this research, a year of trip data and secondary datasets are used to analyse BSS usage in the three case study cities. How land use, socio-economic and network factors influence BSS use was examined using bivariate correlation analysis and through the development of linear regression models for each case study. Bivariate correlations show significant positive association with the number of cafes and restaurants, vicinity to the beach or promenade and the percentage of foreign population at the BSS station locations in all cities. A positive relation with cycling infrastructure is evident in Limassol and Las Palmas de Gran Canaria, but not in Malta, as no cycling infrastructure is present in the island’s conurbation, where the BSS is primarily operational. Elevation has a negative association with BSS use in all three cities. Apart from these similarities, the regression models show more fine-grained results and explain differences in BSS use. The insights from the correlation analysis and regression models can be used to inform policies promoting cycling and BSS use and support sustainable mobility policies in the case study cities and cities with similar characteristics.

Yuchuan Jin, KTH Royal Institute of Technology
Xinwei Ma (230169206@seu.edu.cn), Southeast University
Yufei Yuan, Technische Universiteit Delft
Mingjia He, Southeast University

As an economical, convenient, and eco-friendly travel mode, bike-sharing has improved urban mobility. However, it is often very difficult to satisfy a balanced utilization of shared bikes due to the asymmetric spatio-temporal user demand distribution and the insufficient numbers of shared bikes, docks or parking areas. If we can predict the short-term bike-sharing demand, it will help operating agencies rebalance bike-sharing systems in a timely and efficient way. This study proposes an Attention-based Graph Convolution Long Short-term Memory network (AttGC-LSTM) framework to predict short-term bike-sharing demand at a station level using multi-source data. These data include historical bike-sharing trip data, historical weather data, users’ personal information, and land use data. The proposed model can extract spatio-temporal information of the bike-sharing system, and predict the short-term bike-sharing rental and return demand. We use graph convolutional neural network (GCN) to extract spatial information and adopt long short-term memory network (LSTM) to extract temporal information from the data. The attention mechanism is applied on the spatial dimension to enhance the ability of learning spatial information in GCN. Results show that the proposed model has the highest accuracy compared with several baseline models; the attention mechanism can help improve the model performance; models that include exogenous variables perform better than the models only consider historical trip data. The proposed short-term prediction model can be used to help bike-sharing users better choose routes and to help operators implement dynamic redistribution strategies.

Dillon Fitch (dtfitch@ucdavis.edu), Institute of Transportation Studies (ITS)
Hossain Mohiuddin, University of California, Davis
Susan Handy, University of California, Davis

Many existing studies of bike-share services focus on system dynamics and user characteristics, but less is known about how bike share influences individual-level travel behavior and attitudes of residents. While attitudes toward bicycling have a particularly strong association with bicycling behavior, little is known about how bike share influences the attitudes of residents. In this study, we examine how a bike-share system influenced travel attitudes of residents through a repeat cross-sectional survey conducted before and after the opening of the system. The study focused on one of the largest dock-less electric-assisted bike-share systems (with a small e-scooter share) in the US, the Jump system that opened in the summer of 2018 across three California cities: Sacramento, West Sacramento, and Davis. Results suggest that the bike-share system is likely to have been responsible, at least in part, for more favorable attitudes toward bicycling and less favorable attitudes toward driving. This research demonstrates that the benefits of bike share can go beyond the general use of the system. Bike share can also be seen as an intervention that has widespread psychological effects that reduce known barriers to bicycling.

Chao Qi, Southeast University
Xuewu Chen (chenxuewu@seu.edu.cn), Southeast University
Mingzhuang Hua, Southeast University
Shujie Zheng, Southeast University

With the development of Internet technology and the concern of saving resources and protecting the environment, dockless bike sharing (DBS) gets rapidly prevailing in various cities. However, due to the vicious competition and the blind expansion in the early stage, many enterprises could not make ends meet and went bankrupt one after another. For DBS operators, one of the highest costs comes from the daily rebalancing of bikes. Therefore, this paper attempts to explore whether there is a self-balance in the bike amount of DBS in the long run by analyzing the daily changes of the DBS bike amount. In this paper, the journey data of Mobike in Nanjing for 3 months were utilized. Firstly, according to the existing researches, 4000 clustering centroids were selected as virtual stations. Secondly, 4000 Voronoi diagrams were created by using ArcGIS and intersected with the corresponding circles, whose radii are 500 meters, as service scope of the virtual station. Subsequently, the journey data of 3 months were manipulated by using SQL Server, and the variation curves of the bike amount per day at each station could be obtained. Finally, all stations were divided into four categories (bike-increasing station, bike-decreasing station, bike-fluctuating station, self-balancing station) by clustering. Those where DBS bike amount can maintain stable are the stations that can achieve self-balance. This research investigates the spatial distribution of different kinds of stations, which will benefit DBS enterprises in their daily operation and scheduling.            Keywords: Dockless Bike Sharing, Self-Balance, Journey Data, Virtual Station, Clustering

Nicole Iroz-Elardo, University of Arizona
Kristina Currans, University of Arizona

Objective: With the mass introduction of shared, dockless e-scooter programs, many cities are struggling to understand injury implications. This article systematically documents what is known about e-scooter injuries using emergency department (ED) studies; it also provides recommendations to better understand the health and safety risks of this emerging mode. Methods: A systematic review was performed for all e-scooter articles through November 2019, retaining injury-related articles. In the case where surveillance data and exposure data were available, injury rates were explored. Results: We identified 18 articles, including: five that used surveillance data methods; seven examining all e-scooter injuries from 1-3 hospitals; and six examining a medically-specific subset of those injured. Variation in the reporting structure of data make pooling difficult, but some trends are emerging. Three surveillance studies report an injury rate of 20-25 ED visits per 100,000 trips. Those injured rarely wear helmets, resulting in a high proportion of head injuries. Extremity injuries, including fractures, are also widespread. The profile of injured appears to be a 30-year old male. However, once normalized by exposure data, females, young and older riders may be at higher risk of injury. Comparisons to other modes remain unclear; this is as much a challenge of the exposure data for the other modes as information on e-scooters. Conclusions: Assumptions about comparisons to bicyclists should be more thoroughly examined. Data harmonization and collaboration between vendors, municipalities, and public health departments would improve the quality of data and resulting knowledge about e-scooter safety risk.

Feifei Xin, Tongji University
Yuxuan Luo, Tongji University
Xiaobo Wang, Tongji University

Dockless Bike-sharing System facilitates travelers' transfer process in transit stations, enlarges coverage and promote accessibility of transit stations, hence an increasing number of travelers begin to use Dockless Shared-bike to transfer at metro stations. Consequently, large numbers of Dockless Shared-bike pile up around metro stations and result in a serious deficiency in parking space for bicycles. To solve this problem, it is necessary to explore how do the transfer behavior and transfer demand change after Dockless Bike-sharing System popularizes. In this paper, we design a questionnaire to collect data includes travelers' socio-economic attributes, transfer behaviors characteristics, and other characteristics. A Price Sensitivity Meter (KLP) model is employed to analyze the threshold value of transfer distance of metro stations. Also, a nested logit model is established to explores the determinant factors for using Dockless Shared-bike to transfer at metro stations. Since Shanghai has the largest number of Dockless Bike-sharing System in China, case study is conducted in Shanghai. Results show after the emergence of Dockless Bike-sharing System, the maximum acceptable distance for using bikes to transfer increases by 13.9%, from 2,203 meters to 2,511 meters, which indicates that the service coverage of transit stations expands. Besides, transfer distance is a crucial factor influencing travelers’ choice. Therefore, since the travelers’ acceptable transfer distance for bicycle increase after the Dockless Bike-sharing System popularize, travelers will be more willing to choose to use Dockless Shared-bike as a transfer mode at the metro station, and it will instigate the transfer demand at metro station.

Tian Wen, Tongji University
Yingying Xing (yingying199004@tongji.edu.cn), Tongji University

Recently, the rapid development of bike sharing system (BSS) has dramatically influenced passengers' travel mode. However, whether the relationship between BSS and public transit ridership is competitive or complementary remains unclear. In this paper, a difference-in-difference (DID) model is proposed to figure out the impact of dockless BSS (DBSS) on bus ridership. The data was collected from Shanghai, China, which includes data from automatic fare collection (AFC) systems and automatic vehicle location (AVL) systems, DBSS transaction data and point of interest (POI) data. The research is based on the route-level, and the results indicate that shared bikes have a negative impact on bus riderships (b) regarding all the travel distance, dockless shared bikes lead to a stronger decrease of bus ridership on weekends than on weekdays. (c) considering the travel distance, the negative impact of shared bikes on bus riderships decays gradually with the increase of travel distance. This paper reveals that the travel distance of passengers greatly influences the relationship between DBSS and public transit.

Xin Chen, Southeast University
Zhenliang Ma, Monash University
Guobin Ye, Southeast University
Zhibin Li, Southeast University

The free-floating bicycle sharing system (FFBSS) service is important in the urban mobility ecosystem. Although many studies conducted surveys to understand the usage characteristics of FFBSS service for the general population, few studies focus on the usage behavior of university students. This paper examines the adoption and usage behavior of FFBSS service and identifies the key service attributes driving satisfaction levels among Chinese university students. Undergraduate and graduate university students are an important and novel population, as they are still forming their values and beliefs, and therefore may be more open to engaging in sustainability efforts (e.g. transportation choice). Descriptive analysis is performed to analyze FFBSS service adoption and usage patterns, FFBSS trip characteristics, and the motivation to use FFBSS service. The results show that the socio-demographic information (e.g. age, gender, grades, e-bike/bike ownership, and monthly expense) significantly influence the service adoption and usage frequency. Time-saving and convenience are the two main motivations driving FFBSS service usage. Multivariate regression and importance-performance analysis (IPA) are conducted to identify important FFBSS service attributes impacting user satisfaction levels and identify priority service attributes for improvement. It highlights the importance of the density of service delivery zones, the number of available bicycles, the position accuracy of bicycles, and the good functioning condition of bicycles. The findings provide useful insights into the planning and management of FFBSS services on campus and in cities.

Jinyang Zhang, Southeast University
Xuewu Chen (chenxuewu@seu.edu.cn), Southeast University
Mingzhuang Hua, Southeast University
Wendong Chen, Southeast University

As an environment-friendly travel mode, free-floating bike sharing (FFBS) becomes popular in China recently. It is of great significance to quantitatively evaluate the environmental benefits of FFBS. This paper estimate the carbon dioxide (CO2) emission reduction of FFBS by combining the one-month journey data of Hellobike (one of China's largest FFBS operators) in Nanjing, the Nanjing FFBS survey, and the modal shift data of Nanjing. The life cycle assessment (LCA) was administered to determine the CO2 emission during the life cycle of FFBS, including production, scheduling, maintenance. The results show that about 76.10 kg CO2 per bike is engendered in the life cycle. Besides, the emission reduction benefit generated by the modal shift from other travel modes to shared bikes is approximately 324.95 tons CO2 per day, and the transfer from private cars to bike sharing makes the greatest contribution. The average travel distance of shared bikes is 1.20 kilometer, and the daily turnover rate is 2.57 times per bike in Nanjing. Based on the above findings, it takes about 119 days to strike a balance between the CO2 emission in the life cycle and the emission reduction benefit. Oversupply of bikes will result in less use of each bike, which will further reduce emission reduction benefits. This research will provide a significant help for transportation planners and decision-makers who are committed to the sustainable development of the FFBS system.   Keywords: Free-floating bike sharing, Emission reduction, Life cycle assessment, Modal shift

Jung-Hoon Cho, Seoul National University
Seung-Woo Ham, Seoul National University
Dong-Kyu Kim (dongkyukim@snu.ac.kr), Seoul National University

With the growth of the bike-sharing system, the problem of demand forecasting has become of great importance to the bike-sharing system. As such, this study aims to develop a novel prediction model that enhances the accuracy of the peak hourly demand. Spatio-temporal Graph Convolutional Network (STGCN) is constructed to consider both the spatial and temporal features. One of the model's essential steps is to determine the main component of the adjacency matrix and the node feature matrix. To achieve this, we use data of 131 days from the bike-sharing system in Seoul and conduct experiments on the models with various adjacency matrices and node feature matrices, including the public transit usage. The results indicate that the STGCN models reflecting the previous demand pattern to the adjacency matrix show eminent performance in predicting demand than the other models. The results also show that the model that includes the bus boarding and alighting records is more accurate than the model that contains records of the subway, inferring that buses have a greater connection to bike-sharing than the subway. Based on the results, the usage of STGCN models, previous demand patterns, and public transit data makes it a prominent method for predicting peak demands.

Jianhong Ye, Tongji University
Ruixiang Zhou, Tongji University
Jiahao Bai, Tongji University
Lei Gao, Tongji University
Ruixin Chen, Tongji University

A green and environment-friendly transportation service with both higher speed and flexibility, shared e-bike is guiding the wave of short-medium distance trips. Given the rapid rise of shared e-bikes companies in recent years, and how little is known about their operations, there is pressing public interest in understanding the impact of these transportation-sharing platforms. This research mainly studies the user characteristics, spatiotemporal characteristics of shared e-bikes, and the differences and similarities between shared e-bike and shared bike services. This study finds that shared e-bikes in Shanghai mainly serve middle-aged and young people with higher education. Most of user's frequency of usage is low, no more than twice a week. Similar to the other transportations, it is primarily used for commuting, which is also illustrated by the difference in time distribution pattern and trips between weekends and weekdays. Trips of it are significantly related to the built environment such as the density of commercial, PT station and hotel. Besides, through a comparative analysis with shared bikes, we found that there are significant differences in travel duration, distance and speed between shared e-bikes and shared bikes in Shanghai. The findings from this research contribute to our understanding how shared e-bikes run. Keywords: Shared e-bikes, Spatiotemporal characteristics, Shared bikes, User characteristics

Steven Gehrke (steven.gehrke@nau.edu), Northern Arizona University
Bita Sadeghinasr, Northeastern University
Ryan (Qi) Wang, Northeastern University
Timothy Reardon, Metropolitan Area Planning Council

The recent introduction and proliferation of privately-operated dockless bikeshare systems across North America has caught many public planning agencies, who seek evidence to understand the extent of their adoption by residents in their communities and its impact on existing transportation systems, by surprise. In Greater Boston, where a public dock-based bikeshare system has been in place for nearly a decade, an innovative agreement involving its regional planning agency and one dockless system operator has sought to offer participating municipalities a more predictable introduction via cross-boundary standardization and trip-level data access. Provided with the former, this study seeks to examine the patterns of system use during the first 18 months following the agreement’s inception in addition to the neighborhood-level predictors of access to dockless bikes and their usage. Specifically, this study looks to offer insight into the spatial equity-related impacts of this promising active mobility option for Boston’s suburbs. Of particular interest, our study finds that neighborhood with a higher share of renter-occupied housing and historically disadvantaged populations had less access to dockless bikes yet were predictive of higher bike usage. An undesirable finding that may be addressed by the implementation of policies such as an equitable dockless bikeshare rebalancing scheme.

Xiaodong Qian (xdqian@ucdavis.edu), University of California, Davis
Miguel Jaller, University of California, Davis
Giovanni Circella, University of California, Davis

Bikeshare systems have attracted increased research interest. However, research considering equity issues in the design of these systems is still lacking. To fill this research gap, this study proposes a multi-objective bikeshare station network optimization model addressing equity considerations. Based on a set of candidate station locations, the model estimates the potential demand (i.e., bikeshare trip productions and attractions) and its distribution, and evaluates performance over a set of objectives to find the most equitable distribution of stations. To solve the model, the authors developed a genetic algorithm heuristic to contend with the non-lineal non-continuous nature of the problem objectives. The study uses the Divvy bikeshare system in Chicago as a case study. To validate the results, the authors compares the model solutions with the system’s historic expansion (e.g., new stations added in 2016). Selecting accessibility as the main objective, results in a system with additional stations in disadvantaged areas, consistent with the 2016 system expansion. On the contrary, the revenue maximizing objectives results in a smaller network of stations with fewer accessibility improvements, especially to disadvantaged communities. A sensitivity analysis uncovers the greatest obstacle (i.e., station cost) to extending more stations in disadvantaged areas. More importantly, from the Pareto frontier of solutions, the authors draw policy and planning recommendations, such as the potential need to provide incentives to private service providers to improve accessibility to disadvantaged communities. The findings highlight the importance of considering accessibility and other equity constraints in developing a more inclusive, equitable and sustainable transportation system.

Zhejing Cao (caozhejing1120@163.com), Tsinghua University

While people are embracing shared e-scooters as a promising alternative for short urban travels, some are concerned about its road safety and riding comfortableness. Whether people will continue to choose this new mode can be influenced by how they experience it. In this study, we characterize street-level environment on individuals’ e-scooter trip base as a representation of users’ past riding experience, and examine its impact on their future transport mode choice. In the case study of Singapore Central Area, we characterize the street environment by three dimensions of comfortableness, safety, and attractiveness, and apply Convolutional Neural Network (CNN) to efficiently capture the human-perspective traits in streetscapes. We further measure e-scooter riding experience of street environment by the peaks, averages, and sums for each rider using e-scooter trajectory data, and base on it to model their mode choice in a stated preference survey with the approach of Deep Neural Network with alternative specific utility functions (ASU-DNN). The result of ASU-DNN exhibits much higher prediction accuracy than the random utility model and is interpretable. Experience averages of pavement smoothness, e-scooter lane separation level, street junction density, and street facility density positively affect future e-scooter probability, while experience average of street shading and land use mix have the negatively impact. E-scooter probability is first increasing and then decreasing with the growth in the experience average of streetscape naturalness, road hierarchy, and pedestrian number. Enlarging the gap between experience high peaks and low peaks for street environment boosts future e-scooter probability. However, excessive experience sum will reverse its initial positive impact on e-scooter probability. While shared e-scooter are facing difficulties fitting in current street space which undermines its initial advantage as a flexible transport mode, the results of this study would inform urban and transport planners of how to retrofit street infrastructures and nudge the experience to positively influence future travel choices for micro-mobility.

Natalia Zuniga-Garcia (nzuniga@utexas.edu), University of Texas, Austin
Natalia Ruiz Juri, University of Texas, Austin
Kenneth Perrine, University of Texas, Austin
Randy Machemehl, University of Texas, Austin

This research considers the problem of using trajectory data in the Mobility Data Specification (MDS) standard to conduct meaningful analyses of infrastructure use by e-scooters without compromising personally identifiable information (PII). We assess the integration of e-scooters into the urban infrastructure in Austin, Texas, using trip trajectory data from an e-scooter provider company and infrastructure geographic inventory information. Our analysis uses more than eleven  million location points from approximately 80,000 e-scooter trips made over a year, which accounts for 1.4 percent of the total e-scooter trips made in the city during the same period. Our results suggest that an average e-scooter trip distance is split between sidewalks (18 percent), bike lanes (11 percent), and roadways (33 percent), with 38 percent across other unclassified areas. Furthermore, approximately 60 percent of the roadway trips are made on principal arterials. An analysis of variance suggests that the mean speed of trips made on sidewalks is slightly lower (6 to 8 percent) than on other types of infrastructure, and weekday and AM-peak hours present higher speeds. This study illustrates the potential use of trajectory data to provide insightful analysis to help understand and regulate the use of emerging mobility services on the current urban infrastructure. It also highlights the importance of providing and maintaining geographic urban inventory data. Even though our analyses were conducted using raw data points, we discuss how partially aggregated data without PII could be used to provide similar insights, which can inform the development and extension of data sharing policies.

Haoyun Wang (haoyun.wang@rutgers.edu), Rutgers University
Robert Noland, Rutgers University
Peng Chen, University of South Florida

The COVID-19 pandemic has been unprecedented in its scale and speed, impacting the entire world, and having an impact on metropolitan transportation systems. New York City was especially hard hit in March and April 2020. A mandatory stay-at-home order was instituted, all but essential businesses were ordered closed. In this paper we examine the impact on the Citi Bike system and the NYC subway. Usage patterns during the lockdown is compared to corresponding days in 2019. Controlling for weather patterns we examine the effect of the lockdown and subsequent reopening of economic activity up to the end of June 2020. The results show that both subway ridership and bikeshare usage plummeted initially; bikeshare usage has nearly returned to normal while subway use remains substantially below pre-COVID ridership.

Rebecca Sanders (rebecca@safestreetsresearch.com), Arizona State University
Trisalyn Nelson, Arizona State University

Since 2017, e-scooters have emerged as a viable transportation option across the U.S. and around the world. Yet relatively little is known about e-scooter use, including how they interact with walking and bicycling and how perceptions of benefits and barriers differ across those three modes. This paper presents the results of an online survey of over 1200 university staff (22% response rate) in Tempe, AZ, seeking to understanding these relative differences. We found that e-scooter trips disproportionately replace walking and bicycling trips as compared to car trips. Bicycling and walking are much more likely to be seen as beneficial for exercise or the environment than e-scooting, but e-scooting is more likely than bicycling to be seen as convenient, faster, and better in the heat than walking. E-scooting is also much less likely than bicycling and walking to be seen as impractical due to the need to carry things or go long distances. Enjoyment of each mode is highly significantly related to perceived benefits, whereas perceived safety is highly significantly related to perceived barriers. Despite e-scooters replacing some walking and bicycling trips, they seem to fill a niche for flexible, non-auto travel in high heat areas. E-scooters are also attractive across gender and race/ethnicity lines. However, worries about traffic safety for all modes suggest that cities aiming to encourage non-auto travel need to focus on improving traffic safety. These findings suggest synergies for transportation and public health professionals seeking to increase traffic safety, transportation equity, and travel options in extreme heat.

Javad Jomehpour Chahar Aman (Jjomehpour@smu.edu), Southern Methodist University
Janille Smith-Colin, Southern Methodist University

Shared electric scooters as an emerging mode of transportation have expanded opportunities for urban mobility. Although rapid introduction of these services has altered the transportation ecosystem, little is known about the experience and expectations of e-scooter riders. In this study app store reviews, from two major micro-mobility companies are investigated using machine learning techniques, to identify the factors that influence rider satisfaction. The Latent Dirichlet Allocation (LDA) model is applied to 12,000 rider-generated reviews to identify fourteen factors related to rider satisfaction. These factors cover various topics such as pricing, safety, customer service, refund, payment, app interface, and ease of use, to name a few. Moreover, name-based gender prediction analysis is employed to identify rider gender, and differences in general review content, and level of satisfaction across gender. Findings were validated using logistic regression and the most significant factors predicting rider satisfaction identified. Research findings contribute to the existing literature by demonstrating the use of app store reviews in a transportation mobility study to investigate shared electric scooter user opinions. The development of a method to assess factors contributing to user or rider satisfaction offers the ability to evaluate current and future e-scooter rider needs, and barriers to access and participation. Thus, providing companies, planners, and policymakers the opportunity to employ consistent, effective, and integrated strategies for improving the e-scooter experience and meeting rider expectations.

Adrian Cottam (acottam1@email.arizona.edu), University of Arizona
Xiaofeng Li, University of Arizona
Mohammad Razaur Rahman Shaon, University of Connecticut
Yao-Jan Wu, University of Arizona

With advancements in technology, shared micromobility transportation modes such as bike-sharing systems and E-scooters have continued to grow in popularity across the United States. Understanding the effects of shared micromobility on society, as well as other alternative travel modes can facilitate the development of a sustainable travel alternative for a community. Furthermore, a comprehensive understanding of how one shared micromobility mode can affect other available alternative modes is valuable for decision making and benefit evaluation. In this study, the effects of the introduction of E-scooters with a pre-existing bike-sharing system were evaluated using real-world data. A difference-in-differences regression model was used to estimate the change in the number of bike-sharing trips after the E-scooters were introduced. Furthermore, a temporal and spatial analysis was conducted to evaluate some of the patterns exhibited by bike-sharing users, and how these patterns may have been affected by the introduction of E-scooters. Results indicated that E-scooters caused a reduction in the number of spontaneous trips taken by bike-sharing users. However, it was also found that the number of trips taken by bike-sharing system users with annual passes was increased. Based on travel patterns, it was hypothesized that this increase in bike-sharing usage is due to E-scooters complementing bike-sharing systems, with users combining travel modes. Furthermore, the methods used in this study applied real-world data to achieve results, and are therefore transferable to other mid-size cities to evaluate the impacts of introducing a new shared micromobility mode.

Hossain Mohiuddin, University of California, Davis
Dillon Fitch, Institute of Transportation Studies (ITS)
Susan Handy, University of California, Davis

Bike-share services around the US have attracted considerable ridership, although little is known about the factors influencing an individual’s intention to use this service. In this study, we explore the influence of socio-demographics, bicycling frequency, individual latent attitudes towards bicycling and cars, and the social environment variables on the intention to use bike-share service in different locational contexts. We use cross-sectional survey data collected in 2016 before the launch of the bike-share system in three Californian cities: Sacramento, West Sacramento, and Davis. We apply an Integrated Choice and Latent Variable (ICLV) model to incorporate latent attitude variables with socio-demographic and bicycling frequency variables. The ICLV framework addresses the measurement error and endogeneity bias issues associated with incorporating latent attitude variables in the model. Our findings show that people who are non-white people, employed full-time, younger, and bicycle frequently have a higher intention to use the bike-share service. We also find the pro-bike latent attitude and the social environment around bicycling positively affect intention, while the pro-car latent attitude negatively affects intention. Residents living in Davis, a community with a strong bicycling culture, report lower intention to use bike-share, all else equal, while residents of downtown Sacramento, where the official interest in bicycling is more recent, report greater intention than residents of other areas. These findings raise interesting questions and merit further investigation with data on the actual use of bike-share systems across several cities with varying levels of bicycling infrastructure and culture.

Jungyeol Hong (jyhongsun@gmail.com), University of Seoul
Eunryong Han, University of Seoul
Dongjoo Park, University of Seoul

Shared bicycles are one of the sharing economies for solving complex urban traffic problems, and their demand has been steadily increasing since the introduction of the shared bicycles in Seoul. The demand for shared bicycles is influenced not only by temporal characteristics, but also by various factors such as the characteristics of the city, the environment around the shared bicycle rental station, and physical urban network. Therefore, the primary purpose of this study is to discover factors affecting the demand for shared bicycles and to develop models that predict the demand for each shared bicycle rental station by time as reflecting the influence of these factors. In this study, 263 shared bicycle rental stations in the four main gates in the center of Seoul were classified through the time series clustering analysis, and the demand of each rental station was estimated by time using the random forest method. As a result, it was found that the amount of rental and return before an hour and the temperature and precipitation before an hour were highly significant factors predicting the demand for the next period. Furthermore, it was found that each cluster model considering the characteristics of time-series changes is more accurate than that of models that are not cluster-specific. It is expected that future research will be able to monitor the inventory of bicycles at rental stations and establish strategies for relocation using predicted demand obtained by the framework of the analysis.

Soheil Sohrabi (sohrabi.s@tamu.edu), Texas A&M University, College Station
Alireza Ermagun, Mississippi State University

This study proposes a pattern detection methodology for dynamic bike sharing station traffic prediction using historical traffic and spatiotemporal variables. Particularly, the proposed model is designed to mitigate two operational challenges of bike sharing systems: developing a short-term trip advisory system and scheduling bike redistribution across the system. We develop the model on the Washington DC Capital Bikeshare data to predict bike sharing station traffic for both short and long terms ranging from 15-minute to 4-hour horizons. The results show that the proposed model can predict the bike sharing station traffic in 15-minute intervals with higher precision for short-term predictions and lower precision for long-term predictions. We highlight that the spatiotemporal variables help improve the prediction accuracy of the model. Temporal characteristics contribute more than spatial characteristics in short-time predictions. The contribution, however, is flipped for long-time horizons. The proposed models have the capacity to estimate bike sharing traffic for both short and long time horizons in less than 20 seconds, which illustrates (1) the practicality of the models in dynamic bike sharing traffic prediction, and (2) the potential of the proposed model to be updated in real-time and incorporate the most recent observations into predictions.

Lori Palaio, University of South Florida
Tung Vo, University of South Florida
Michael Maness, University of South Florida
Robert Bertini, Oregon State University
Nikhil Menon, University of South Florida

Bikeshare provides important first mile last mile, commuting, circulation, and sightseeing options in many cities. Bikeshare can also be healthy and convenient for users. Throughout the year, holidays occur which change typical bikeshare activity patterns. Existing literature shows mixed results relating to the ridership impacts of holidays, as some research shows that these days may result in higher ridership, while others show no effect. Because of variations in system locations and modeling methods, it is difficult to determine the reasons for these mixed results. To control for these aspects, this project consists of a multi-city study of the effect of holidays on system-level ridership using a log-linear regression model with robust standard errors. The results show the impacts of holidays on bikeshare system ridership for different user types among systems in the Washington D.C., Chicago, Boston, Los Angeles, and Minneapolis metro areas. Several hypotheses are built and tested for examining the expected effects of holidays on bikeshare usage. A major finding from this study is that federal holidays negatively affect member ridership and positively affect non-member ridership. It was also found that different federal holidays have dissimilar effects on total ridership. These findings could be useful for bikeshare agencies to plan, reposition fleet, and improve system operation.

Farzana Mehzabin Tuli, University of Arkansas, Fayetteville
Suman Mitra, University of Arkansas, Fayetteville

The rapid popularity growth of shared e-scooter creates the necessity of understanding the determinants of shared e-scooter usage. This paper estimates the impacts of temporal variables (weather data, weekday/weekend, and gasoline price), socio-demographic, built environment, and neighborhood characteristics on the shared e-scooter demand by using four months (June- October 2019) period of data from the shared e-scooter pilot program in Chicago. The study employs a random effect negative binomial (RENB) model that effectively models shared e-scooter trip count data with over-dispersion while capturing serial autocorrelation in the data. Results of the RENB model reveal that the important determinants (significant in both models) that contribute to increases in shared e-scooter demand in Chicago due to a unit change in the respective variable are, in order of IRR-estimated magnitude (average of origin and destination models): i) income (higher-income: 171%, medium-income: 88%; baseline: low-income), ii) gasoline price (119.5%), iii) land-use mix (54%), iv) percentage of zero car households (30.5%), v) parking rate (19.5%), vi) census tracts with one train station (17.5%; baseline: no station). On the other hand, the significant determinants that are found to decrease the shared e-scooter usage are: i) precipitation (-16.5%), ii) the number of homicide records (-8%). The rest of the variables either are not significant in both models or contribute to zero to less than (±)5% changes in shared-scooter demand in Chicago. The findings of this study will help planners and policymakers make decisions and policies related to shared e-scooter services.