•A comprehensive comparison between machine learning and logit models is provided.•Random forest model has much higher predictive accuracy compared to multinomial logit model and mixed logit ...model.•Machine learning and logit models agree on many aspects of the behavioral outputs.•Applying a standard approach to generate marginal effects and arc elasticities for random forest lead to unreasonable estimates.•Random forest captures nonlinear relationships between the input features and choice outcomes.
Some recent studies have shown that machine learning can achieve higher predictive accuracy than logit models. However, existing studies rarely examine behavioral outputs (e.g., marginal effects and elasticities) that can be derived from machine-learning models and compare the results with those obtained from logit models. In other words, there has not been a comprehensive comparison between logit models and machine learning that covers both prediction and behavioral analysis, two equally important subjects in travel-behavior study. This paper addresses this gap by examining the key differences in model development, evaluation, and behavioral interpretation between logit and machine-learning models for mode-choice modeling. We empirically evaluate the two approaches using stated-preference survey data. Consistent with the literature, this paper finds that the best-performing machine-learning model, random forest, has significantly higher predictive accuracy than multinomial logit and mixed logit models. The random forest model and the two logit models largely agree on several aspects of the behavioral outputs, including variable importance and the direction of association between independent variables and mode choice. However, we find that the random forest model produces behaviorally unreasonable arc elasticities and marginal effects when these behavioral outputs are computed from a standard approach. After the introduction of some modifications that overcome the limitations of tree-based models, the results are improved to some extent. There appears to be a tradeoff between predictive accuracy and behavioral soundness when choosing between machine learning and logit models in mode-choice modeling.
This article describes the methodology used to study the quality of service desired by users of a public transport system. The desired quality is different from the perceived quality because it does ...not represent the daily experiences of the users, but rather what they desire, hope for or expect from their public transport system. This is why it is important to study the desired quality, knowledge of which gives local authorities the background information for personalised marketing policies based on the user’s requirements rather than their daily perceptions. The methodology goes through several stages, such as the use of focus groups to choose the most important variables for the users, the design and use of unlabelled stated preferences surveys and the calibration of discrete choice models. All of these help determine the weight of the most relevant variables. The analysis is carried out with different categories of users and potential users (those people not currently using public transport).
Waiting time, cleanliness and comfort are shown to be the public transport variables that users most valued, but the degree to which they are valued varies according to the category of user. Variables such as driver kindness, bus occupancy and journey time are generally given less weight. The first two vary little by user category, but some variability appears for journey time.
For potential users the more important variables when defining expected quality from public transport are waiting time, journey time and above all, level of occupancy. They consider the other variables to be of little importance when defining an efficient public transport service.
In order to improve service quality and attract more passengers to public transport in general, the application of this methodology provides the authorities and operating companies with useful information to plan personalised marketing policies specifically directed at different categories of users and potential users of public transport.
►The quality of a public transport system is covered by many factors, such as considerations relative to comfort and safety within the vehicle, the time taken to cover the routes and the convenience and existence of any supporting infrastructure ►The desired quality is different from the perceived quality because it does not represent the daily experiences of the users, but rather what they desire, hope for or expect from their public transport system. ►This paper offers a new point of view by introducing a new idea of quality: the quality the user and potential user desires. ►Desired quality is shown to vary with user category. ►In order to improve service quality and attract more passengers to public transport in general, the application of this methodology provides the authorities and operating companies with useful information to plan personalised marketing policies specifically directed at different categories of users and potential users of public transport.
In the context of first/last mile, the bicycle-transit intermodality could lessen accessibility problems of stations and stops, reducing the need for feeder services. Although extensively addressed ...in developed countries, bicycle-transit is less studied in Global South metropolises, where distinct usage patterns are observed. Thus, this research aimed to model the choice behavior of the low-income population for bicycle-transit integration in Fortaleza - Brazil, a large Latin American city with a deployed bike-sharing system focused mainly on bike-bus integration. A Stated Preference (SP) survey was conducted addressing socioeconomic and trip characteristics, as well as policy variables such as bicycle parking and cycleways/lanes. SP data allowed the estimation of nested and mixed logit models representing choice behavior, and sample segmentation was used to identify heterogeneity among different groups of individuals. The results show that investing in bicycle infrastructure could stimulate bike-bus integration in Fortaleza; however, their relative importance depends on the analyzed individual strata. Findings also indicate the influence of public and road safety issues, as well as gender, income, and trip characteristics such as purpose and access distance.
•A best-worst model is estimated to explore the role of AV attributes in customers’ adoption decision.•This model can differentiate between the direct impacts of the attributes and the impact of the ...attribute levels.•Vehicle price, exclusive lane, and liability are found as the most important attributes in market penetration of AVs.
Autonomous mobility is one of the rapidly evolving aspects of smart transportation which carries the potential of reshaping both demand and supply sides of transportation systems. While understanding public opinions about autonomous vehicles (AVs) is a compelling step towards their successful implementation, still little is known about to which extent people will embrace this new technology and how the vehicle features will affect their adoption decision. This study presents a new approach for modeling the adoption behavior of fully AVs using the profile-case best-worst scaling model. In this approach, an AV profile which is characterized in terms of the main vehicle attributes and their associated levels is presented to the decision maker and he/she is asked to select the most and the least attractive attributes. Further, a binary adoption question at the end of the choice task inquires if the respondent is willing to purchase the described AV. Utilizing this method, we can recognize the difference between the intrinsic impacts of the vehicle attributes and the impact of the attribute levels on the adoption decision. Results of the analysis indicate that people are much more sensitive to the purchase price and incentive policies such as taking liability away from the “driver” in case of accidents and provision of exclusive lanes for AVs compared to other factors such as fuel efficiency, safety, or environmental friendliness. Further, it is found that millennials with higher income, those who live in the downtown area, and the majority of people who have experienced an accident in the past have greater interests in adopting this technology.
Uneven utilization of airport clusters is a widespread problem, yet China is still planning to build more airports. Intermodal Passenger Transport Service (IPTS), the most typical of which is the ...Air-Rail Integration Service (ARIS), is one new way to solve this problem. Although some service products have been derived, intermodal travel services are still in their infancy due to problems such as long transfer time and inconvenient connectivity arising from the fact that those products fail to take into account the complexity of transfers between multiple hubs, resulting in a low penetration rate of intermodal services. Therefore, this study constructs a universal Hybrid Choice Model (HCM) framework that includes mode choice and city choice based on an online stated preference (SP) survey about multi-hub intermodal transportation in the Beijing-Tianjin-Hebei urban agglomeration, and introduces travelers' concerns about travel security, comfort, convenience and reliability as attitudinal variables, which expands the research on IPTS. Then the willingness to pay (WTP) for travel service is evaluated to demonstrate intermodal travelers' preference for shorter transfer time, transferring in the metropolis, more reserve time for air than rail, and the impact mechanisms of luggage through-handling service and mutual recognition mechanism of security checks. Last but not least, through scenario testing, we find that the excessive concern for security and comfort among high-educated and high-income passengers with a low tolerance for transfer time will reduce the market share of air-rail. These findings can assist relevant departments and businesses in making more humane plans for intermodal travel and infrastructure construction.
•Multi-hub intermodal travel is considered for balancing airports' utilization.•A universal HCM framework covering mode choice and city choice is constructed.•Four attitudinal variables are introduced to reveal travelers' heterogeneity.•Differences in intermodal travelers' concerns between China and Europe are showed.•The WTP of luggage through-handling service is related to the security checks.
•Results from a survey that focused on public attitudes toward emerging mobility options are presented.•Eleven latent factors that represent various aspects of attitudes were identified.•The ...underlying patterns of attitudes were discussed.•The correlations between the latent attitudes and the observed covariates were examined.
This paper presents a comprehensive analysis of people’s attitudes toward shared mobility options and autonomous vehicles (AVs), with a focus on the underlying patterns and potential determinants. A stated preference (SP) survey was designed and implemented in the U.S. Four sets of questions were included in the questionnaire, each focused on one unique aspect of user attitudes, including a) preferences for mobility options and lifestyle (such as overall view of driving, factors in mode choice decisions and technology engagement), b) perceived benefits and concerns of shared mobility option, c) reasons toward or against private vehicle ownership, and d) motivations for and desired features of AVs. A structural equations model was developed to identify latent attitudinal factors and examine the correlations between the latent attitudes (as the endogenous variables) and the observed covariates (including the socio-economic and demographic characteristics, and users’ current mobility profile, such as mode use frequency, travel distance, and trip fare). The model identified eleven latent factors that represent various aspects of attitudes toward AVs and shared mobility options. The findings could be used by policymakers and Transportation Network Companies (TNCs) to a) recognize the users’ latent attitudes, b) understand the underlying patterns of attitudes, c) implement plans and policies more efficiently, d) guide or influence users’ perceptions, and e) enhance travel behavior models. This study lays the foundation for further analysis on understanding user acceptance and adoption of these emerging mobility options, which is essential to estimate the likelihood and magnitude of behavior shifts in the era of automated, connected and shared mobility.
•Flexible, demand-adaptive transit services are intended for low-density areas.•These services provide both local circulation and access to mileposts such as regional rail stations.•Current mode ...usage has strong influence on likelihood of using flexible transit service.•Waiting at one’s origin is considerably less onerous than waiting at transit stop.•Results inform the service design process for flexible transit and mobility services.
This paper assesses the demand for a flexible, demand-adaptive transit service, using the Chicago region as an example. We designed and implemented a stated-preference survey in order to (1) identify potential users of flexible transit, and (2) inform the service design of the flexible transit mode. Multinomial logit, mixed-logit, and panel mixed-logit choice models were estimated using the data obtained from the survey. The survey instrument employed a dp-efficient design and the Google Maps API to capture precise origins and destinations in order to create realistic choice scenarios. The stated-preference experiments offered respondents a choice between traditional transit, car, and a hypothetical flexible transit mode. Wait time, access time, travel time, service frequency, cost, and number of transfers varied across the choice scenarios. The choice model results indicate mode-specific values of in-vehicle travel time ranging between $16.3 per hour (car) and $21.1 per hour (flexible transit). The estimated value of walking time to transit is $25.9 per hour. The estimated value of waiting time at one’s point of origin for a flexible transit vehicle is $11.3 per hour; this value is significantly lower than the disutility typically associated with waiting at a transit stop/station indicating that the ‘at-home’ pick-up option of flexible transit is a highly desirable feature. The choice model results also indicate that respondents who use active-transport modes or public transit for their current commute trip, or are bikeshare members, were significantly more likely to choose flexible and traditional transit than car commuters in the choice experiments. The implications of these and other relevant model results for the design and delivery of flexible, technology-enabled services are discussed.
Trade Uber for the Bus? Dong, Xiaoxia
Journal of the American Planning Association,
04/2020, Letnik:
86, Številka:
2
Journal Article
Recenzirano
Problem, research strategy, and findings: Few studies have examined ride-hail users' individual preferences between ride-hail and transit. Based on a survey of ride-hail users in the Philadelphia ...(PA) region, I examine who uses ride-hail and investigate ride-hail users' willingness to use ride-hail versus transit. My results suggest that more than one-quarter of respondents replaced transit with ride-hail in their last ride-hail trips. Mixed logit regression analysis based on stated preference choice experiments indicate that higher-income respondents and respondents over 30 years old are increasingly willing to choose ride-hail over transit, even though their actual ride-hail usage is lower than that among lower-income and younger respondents. Results also show that female respondents are more willing to choose ride-hail over transit than male respondents and less frequent transit users are more likely to choose ride-hail than frequent transit users. Higher cost and longer trip duration are significant deterrents for travel by either mode. Respondents consider the time spent on walking to and from transit more burdensome than in-vehicle travel time and wait time for transit and ride-hail. They consider waiting for ride-hail less burdensome than waiting for transit. Survey sampling and design limitations provide lessons for future ride-hail studies.
Takeaway for practice: Practitioners should ensure convenient, affordable travel options for lower-income residents, who are more frequent but less willing ride-hail users than higher-income residents. Female respondents' safety concerns about transit should urge transit agencies to recognize female transit riders' travel needs. The relationship between age and willingness to use ride-hail reminds planners to anticipate greater substitution of ride-hail for transit as the more tech-savvy generation starts entering their 30s. Last, fare reduction alone may not be enough to prompt ride-hail users to switch to transit. Service improvements that shorten the overall trip duration are imperative to make transit more attractive.
After the COVID-19 pandemic, a decrease in travel demand due to the mass transport characteristics of public transit was inevitable, and the impact of the pandemic has increased the need for new ...transport policies. This study aims to propose transport policies for public transit through comparison by quantitatively measuring passengers' crowding impedance before and after the COVID-19 pandemic. The crowding impedances are evaluated through a random parameter mixed logit model from the two surveyed data to compare the behavioral differences before and after the COVID-19 pandemic. The behavioral differences are compared using crowding multipliers, and, accordingly, the results show that crowding impedances after the COVID-19 pandemic are about 1.04 ~ 1.23 times higher than before the COVID-19 pandemic. Transport policies for transit policymakers and operators are proposed to cope with the crisis in public transportation caused by an infectious disease.
The technological innovation has gradually turned UAV delivery services in urban areas into a reality. As crucial stakeholders, urban residents have a growing demand for goods delivery, and they are ...in the process of understanding this new technology. Understanding their acceptance and quantifying their preferences can effectively promote the development of UAV delivery systems. While scholars have conducted extensive research on this topic, there remains a lack of detailed analysis regarding consumer heterogeneity and specific service experiences, especially in China, which is one of the largest potential markets for urban air mobility and UAV applications. This study addresses pertinent questions, including in high-density urban areas, the potential user characteristics of UAVs for instant delivery, the types of goods users prefer UAVs to transport, and how users prefer goods to be handed over to them. Conducted in Chengdu, a major city in southwest China, the survey collected 2,008 validated responses, encompassing potential users of urban air delivery. Employing a discrete choice model (DCM) for quantitative analysis, market share and willingness to pay were derived based on field data, with the identified issues designed as attribute variables integrated into the model. The findings reveal that under suitable services, the UAV market share can reach 21.2%, while electric bike and car deliveries persist as the mainstream in on-demand delivery, constituting 51.8% and 27.0%, respectively. Notably, consumers’ choices are significantly influenced by their socioeconomic status and key indicators. As the cost of UAVs increases relative to electric bike and car deliveries, respondents tend to prefer the electric bike, indicating that UAVs are not the preferred substitute. However, when the price falls below 9.5 CNY, UAVs become more appealing than cars. The identification of five latent classes reflects distinctly different user attitudes towards delivery services in China. The analysis of key attribute elasticity guides adjustments in delivery efficiency and pricing, aligning with market conditions. These results elucidate the intriguing psychology of Chengdu consumers when embracing this new technology. For actively accepting consumers, the hope is for UAV delivery to offer a more convenient service experience, with customized services especially between the arrival of UAVs and pick-up emerging as a distinctive feature. We also discusses regional heterogeneity and implications of the pandemic. As an effective supplement to existing studies, we posit that when UAVs meet customer demands for affordability and high-quality service, they will emerge as the primary force in Chengdu’s future urban air market.
•UAV instant delivery depends on goods type and customers’ socioeconomic status.•Users prefer UAV with lower delivery prices, short waits, and optimized goods access.•UAV delivery should evolve towards a customized and tiered approach to add value.•UAVs are more attractive when the delivery cost ranges between 0 and 9.5 CNY.