2021

Yueshuai He, University of California, Los Angeles
Joseph Chow, New York University

Mobility-as-a-Service systems become increasingly important in the context of smart cities, with challenges arising for public agencies to obtain data from private operators. Only limited mobility data were provided to city agencies, which are not enough to support the decision-making of agencies. This study proposed an entropy maximizing gravity model to predict origin-destination patterns of both passenger and mobility fleet with only partial operator data. An iterative balancing algorithm was proposed to efficiently reach the entropy-maximization state. With different trip length distributions data available, two calibration applications were discussed and validated with a small-scale numerical example. Tests were also conducted to verify the applicability of proposed model and algorithm to large-scale real data of Chicago TNC. Both shared-ride and single-ride trips were forecasted based on the calibrated model, and the prediction of single-ride has a higher level of accuracy.

Shuqing Kang, University of Texas, Austin
Aupal Mondal, University of Texas, Austin
Aarti Bhat, Pennsylvania State University
Chandra Bhat (bhat@mail.utexas.edu), University of Texas, Austin

Pooled mobility services hold substantial promise as a means to provide better accessibility to those who may find it difficult to drive themselves, while also promoting sustainable transportation efforts. In this paper, we develop a joint revealed preference-stated preference model for the choice between pooled versus private ride-hailing that (a) accommodates a suite of individual-level socio-demographics, individual-level psycho-social attributes, built environment variables, and trip-level variables, and (b) explicitly recognizes the importance of considering familiarity with pooled ride-hailing (RH) as an integral element of the pooled RH choice process. Our results underscore the value of using psycho-social latent constructs in the adoption of current and emerging mobility services. Women, older adults, and non-Hispanic/non-Latino Whites have a low propensity to choose the pooled RH mode, while employed individuals, highly educated individuals, and those living in high density urban areas have a high propensity. Overall, the average VTT estimate is $27.80 per hour for commute travel, $19.40 per hour for shopping travel, and $10.70 per hour for leisure/social travel, while the willingness to pool (that is, the willingness to pay to not pool a ride or WTS) averages about 62 cents for commute travel, $1.70 for shopping travel and $1.32 for leisure travel. These estimates can be used by TNCs and cities to consider new integrated pooled RH-fixed transit service designs, position traffic congestion alleviation strategies and new mobility services, and customize information campaigns to promote pooled RH mode use.

Wenwen Zhang (wenwenz3@vt.edu), Virginia Polytechnic Institute and State University (Virginia Tech)
Ralph Buehler, Virginia Polytechnic Institute and State University (Virginia Tech)
Andrea Broaddus, Ford
Ted Sweeney, Spin

E-scooters are an innovative means of transportation that meet the travel demand of many travelers and are well-suited to replace trips shorter than 2 miles, which account for 36% of all trips in the U.S. A lack of understanding of user preferences regarding routing and infrastructure make it difficult for policymakers and planners to adapt the existing transportation infrastructure to accommodate these emerging travel demands. This study develops an e-scooter route choice model to reveal riders’ preferences for different types of transportation infrastructure, using revealed preferences data. The data were collected using Global Positioning System (GPS) units installed on e-scooters operating on Virginia Tech’s campus in September and October 2019. The route choice model was developed using the Recursive Logit (RL) Model, which overcomes the limitations of conventional discrete choice models by adopting unconstrained dynamic choice sets. RL models are applied to 2,000 randomly sampled and post-processed e-scooter trajectories. The model results suggest the infrastructure preferences of e-scooter riders are quite similar to cyclists’ preferences. E-scooter riders are willing to travel longer to use bikeways, multi-use paths, tertiary roads, and one-way roads, and tend to avoid routes interrupted by stairs. Like cyclists, e-scooter users also prefer shorter and simpler routes (i.e., straighter routes with less left and U-turns). Finally, slope is not a determinant for route choice for e-scooter riders, likely because e-scooters are powered by electricity.

Jiangbo Yu, AECOM
Michael Hyland, University of California, Irvine

Modeling and predicting users’ preferences and response times (RTs) under different pricing schemes and information frames are critical in evaluating and improving Mobility-on-Demand (MOD) services. Operators’ information provision, delay, and vehicle allocation strategies influence users’ preferences and RTs, which, in turn, affect which operational and information provision strategies are optimal. Evidence shows that preferences and RTs are sensitive to precedent decisions, information frames, risk, and pressure levels. These interplaying factors are challenging to capture using a traditional discrete choice modeling framework. We propose a generalized diffusion model based on Decision Field Theory (DFT), multi-attribute Prospect Theory (PT), and Random Utility Model (RUM) to address these challenges. A case study applies the model in the context of ordering Shared-Use Autonomous Vehicle Mobility Services (SAMS). The sensitivity test explores various inputs and parameters such as initial waiting time estimate, updated waiting time estimate, time pressure, loss aversion, and value-of-time. The proposed model provides value to MOD operators’ information provision strategies and urban stakeholders’ high-resolution impact analysis of MOD-related pricing policies and regulations. The modeling framework is capable of extending to other applications where multiple sub-decisions are needed to complete one single trip decision under information update, framing, risk, and time pressure.

Muntahith Orvin, University of British Columbia
Jashan Bachhal, University of British Columbia
Mahmudur Fatmi, University of British Columbia

This study presents the findings on modeling the demand of shared e-scooter services (SES); specifically, the spatio-temporal variation of SES demand. A zero-inflated negative binomial (ZINB) model is developed using count of trip origin data at the dissemination-level from Kelowna, Canada. The motivation for adopting the ZINB model is the presence of excess zeros in the count data. ZINB has two components: the zero-inflated component accounts for excess zeros, and the count component accounts for the over-dispersion characteristics of data resulting from excess zeros. In addition to the ZINB, several other count models including the hurdle models are estimated. The goodness-of-fit measures suggest that the ZINB model outperforms other methods. The model results confirm the effects of temporal, weather, transportation infrastructure, land use, and neighborhood characteristics. For example, the count model results reveal that SES demand is more likely to be higher during summer, mid-day of weekend, afternoon of weekday, and days without rainfall. Furthermore, higher e-scooter index, higher density of cycle track, heterogeneous land use, urban centers, lower elevation, and neighborhoods with higher density of hotels and younger population might induce higher demand. The results of the zero component of the model are consistent with the findings revealed by the count component. The model is validated using a hold-out sample, and the validation results confirm that the prediction performance of the model is reasonably satisfactory. The findings of this study provide important insights regarding when and where the demand is higher, which will assist in effective e-scooter supportive policy-making.