Latent Class Choice Model Research Articles

Understanding California wildfire evacuee behavior and joint choice making

For evacuations, people must make the critical decision to evacuate or stay followed by a multi-dimensional choice composed of concurrent decisions of their departure time, transportation mode, route, destination, and shelter type. These choices have important impacts on transportation response and evacuation outcomes. While extensive research has been conducted on hurricane evacuation behavior, little is known about wildfire evacuation behavior. To address this critical research gap, particularly related to joint choice-making in wildfires, we surveyed individuals impacted by the 2017 December Southern California Wildfires (n = 226) and the 2018 Carr Wildfire (n = 284). Using these data, we contribute to the literature in two key ways. First, we develop two latent class choice models (LCCMs) to evaluate the factors that influence the decision to evacuate or stay/defend. We find an evacuation keen class and an evacuation reluctant class that are influenced differently by mandatory evacuation orders. This nuance is further supported by different membership of people to the classes based on demographics and risk perceptions. Second, we develop two portfolio choice models (PCMs), which jointly model choice dimensions to assess multi-dimensional evacuation choice. We find several similarities between wildfires including a joint preference for within-county and nighttime evacuations and a joint dislike for within-county and highway evacuations. Altogether, this paper provides evidence of heterogeneity in response to mandatory evacuation orders for wildfires, distinct membership of populations to different classes of people for evacuating or staying/defending, and clear correlation among key wildfire evacuation choices that necessitates joint modeling to holistically understanding wildfire evacuation behavior.

Potential of on-demand services for urban travel

On-demand mobility services are promising to revolutionise urban travel, but preliminary studies are showing they may actually increase total vehicle miles travelled, worsening road congestion in cities. In this study, we assess the demand for on-demand mobility services in urban areas, using a stated preference survey, to understand the potential impact of introducing on-demand services on the current modal split. The survey was carried out in the Netherlands and offered respondents a choice between bike, car, public transport and on-demand services. 1,063 valid responses are analysed with a multinomial logit and a latent class choice model. By means of the latter, we uncover four distinctive groups of travellers based on the observed choice behaviour. The majority of the sample, the Sharing-ready cyclists (55%), are avid cyclists and do not see on-demand mobility as an alternative for making urban trips. Two classes, Tech-ready individuals (27%) and Flex-ready individuals (9%) would potentially use on-demand services: the former is fairly time-sensitive and would thus use on-demand service if they were sufficiently fast. The latter is highly cost-sensitive, and would therefore use the service primarily if it is cheap. The fourth class, Flex-sceptic individuals (9%) shows very limited potential for using on-demand services.

Traveller behaviour in public transport in the early stages of the COVID-19 pandemic in the Netherlands.

With a few exceptions, public transport ridership around the world has been hit hard by the COVID-19 pandemic. Travellers are now likely to adapt their behaviour with a focus on factors that contribute to the risk of COVID-19 transmission. Given the unprecedented spatial and temporal scale of this crisis, these changes in behaviour may even be sustained after the pandemic. To evaluate travellers’ behaviour in public transport networks during these times and assess how they will respond to future changes in the pandemic, we conduct a stated choice experiment with train travellers in the Netherlands at the end of the first infection wave. We specifically assess behaviour related to three criteria affecting the risk of COVID-19 transmission: (i) crowding, (ii) exposure duration, and (iii) prevalent infection rate. Observed choices are analysed using a latent class choice model which reveals two, nearly equally sized latent traveller segments: ‘COVID Conscious’ and ‘Infection Indifferent’. The former has a significantly higher valuation of crowding, accepting, on average 8.75 min extra waiting time (average total travel time in the choice scenarios was about 40 min) to reduce one person on-board. Moreover, this class indicates a strong desire to sit without anybody in the neighbouring seat and is quite sensitive to changes in the prevalent infection rate. By contrast, the Infection Indifferent class has a value of crowding (1.04 waiting time minutes/person) that is only slightly higher than pre-pandemic estimates and is relatively unaffected by infection rates. We find that older and female travellers are more likely to be COVD Conscious while those reporting to use the trains more frequently during the pandemic tend to be Infection Indifferent. Further analysis also reveals differences between the two segments in attitudes towards the pandemic and self-reported rule-following behaviour. We believe that the behavioural insights from this study will not only contribute to better demand forecasting for service planning but will also inform public transport policy decisions aimed at curbing the shift to private modes.

Train travel in corona time: Safety perceptions of and support for policy measures.

To minimize the risk of becoming infected by the Coronavirus while traveling by train, the national government and the Dutch railways' operator (NS) in the Netherlands have taken several policy measures. These involve that passengers have to wear masks and guidelines are issued for working at home and teaching online. In addition, other policy measures, such as introducing a reservation system, were considered. To examine to what extent train travelers support policy measures and how these change their perception of becoming infected while traveling by train, a stated preference experiment is conducted. Respondents were asked to evaluate various combinations of policy measures, both whether they consider it safe to travel by train under the stated conditions and whether they would vote in favor of the policy package in a referendum. To analyze the data, a mediation choice model is developed, which allows disentangling the direct effect of the policy measures on support and the indirect effect mediated by infection safety perception. To illustrate this, the results show that implementing the policy measure teaching on campus with later starting times would decrease travelers’ infection safety perception and therefore indirectly decrease its support. However, the positive direct effect on support suggests that travelers like this option better than teaching online, the guideline that applied at the time of data collection. The direct and indirect effects cancel each other out, indicating that this alternative policy measure would not count on more support than the guideline teaching online. Furthermore, this paper examines the heterogeneity in the support for policy measures by presenting and discussing the results of a Latent Class Choice Model. Amongst others, the results reveal that one class strongly supports the policy measure reservation system, while another class stongly opposes this measure, suggesting that implementing this measure is not trivial as suggested by its moderate effects at the aggregate level.

Willingness of Hurricane Irma evacuees to share resources: a multi-modeling approach

ABSTRACT Recent technological improvements have expanded the sharing economy (e.g. Airbnb, Lyft, and Uber), coinciding with a growing need for evacuation resources. To understand factors that influence sharing willingness in evacuations, we employed a multi-modeling approach using three model types: (1) four binary logit models that capture sharing scenario separately; (2) a portfolio choice model (PCM) that estimates dimensional dependency, and (3) a multi-choice latent class choice model (LCCM) that jointly estimates multiple scenarios via latent classes. We tested our approach by employing online survey data from Hurricane Irma (2017) evacuees (n=368). The multi-model approach uncovered behavioral nuances undetectable with one model. For example, the multi-choice LCCM and PCM models uncovered scenario correlation and the multi-choice LCCM found three classes – transportation sharers, adverse sharers, and interested sharers – with different memberships. We suggest that local agencies consider broader sharing mechanisms across resource types and time (i.e. before, during, and after evacuations).

Gaussian process latent class choice models

We present a Gaussian Process – Latent Class Choice Model (GP-LCCM) to integrate a non-parametric class of probabilistic machine learning within discrete choice models (DCMs). Gaussian Processes (GPs) are kernel-based algorithms that incorporate expert knowledge by assuming priors over latent functions rather than priors over parameters, which makes them more flexible in addressing nonlinear problems. By integrating a Gaussian Process within a LCCM structure, we aim at improving discrete representations of unobserved heterogeneity. The proposed model would assign individuals probabilistically to behaviorally homogeneous clusters (latent classes) using GPs and simultaneously estimate class-specific choice models by relying on random utility models. Furthermore, we derive and implement an Expectation-Maximization (EM) algorithm to jointly estimate/infer the hyperparameters of the GP kernel function and the class-specific choice parameters by relying on a Laplace approximation and gradient-based numerical optimization methods, respectively. The model is tested on two different mode choice applications and compared against different LCCM benchmarks. Results show that GP-LCCM allows for a more complex and flexible representation of heterogeneity and improves both in-sample fit and out-of-sample predictive power. Moreover, behavioral and economic interpretability is maintained at the class-specific choice model level while local interpretation of the latent classes can still be achieved, although the non-parametric characteristic of GPs lessens the transparency of the model.

Investigating the effectiveness of COVID-19 pandemic countermeasures on the use of public transport: A case study of The Netherlands

During the COVID-19 pandemic, public transport in many cities faces dramatic reduction of passenger demand. Various countermeasures such as social distancing and in-vehicle disinfection have been implemented to reduce the potential risks concerning infection, the effectiveness in promoting the use of public transport however remains unclear. Unlike the usual situation where time and cost are the main factors affecting travel decisions, the uncertainty hiding behind the behavior change of public transport users in a pandemic might be greatly affected by the control measures and the perception of people. This paper therefore aims to examine the effects of COVID-19 related countermeasures implemented in public transport on individuals' travel decisions. We explore the extent to which do policy countermeasures influence different groups of people on the use of public transport. An error component latent class choice model was estimated using the data collected in the Netherlands. Results show that the restrictions policy lifted by the Dutch central government have significant effect on individuals' transportation mode choice decision during the pandemic. The related measures adopted by the public transport sector, by contrast, present different effects on different people. The older and highly educated people are more susceptible to enforcement measures, whereas young and single Dutch citizens are more accessible to non-compulsory measures. Moreover, compared with other private modes, public transport is generally identified as a riskier option, and the average willingness to travel descends. Findings of this study are helpful for the authorities in designing and promoting effective policies in the context of pandemics.

Prioritize your heat pump or electric vehicle? Analysing design preferences for Direct Load Control programmes in Swiss households

Direct Load Control (DLC) of heat pumps and electric vehicles (EVs) could offer great potential to decarbonize the distribution grid network by aligning demand with intermittent renewable technologies as well as avoiding congestion during peak times. However, little is known about consumer preferences for different designs of a DLC programme and their reasonings. To address this gap, this study investigates the preferences for the design of DLC programmes for heat pumps and EVs, using a discrete choice experiment on a sample of 556 respondents residing in Switzerland. We then propose a theoretical framework to provide insights into the underlying reasons such as social-psychological factors and perceptions for those preferences. We applied a multinomial logit analysis to elicit the preferences and found that overall sample was more sensitive to financial incentives for the DLC for heat pumps whereas more sensitive to the overriding option for the DLC for EVs. We then estimated a latent class choice model to explore the heterogeneity in preferences, identifying three distinct classes that vary in their preferences of different attributes. The model suggests that the preferences for the design are influenced mainly by perceived concerns related to being too dependent on the DLC programmes, trust in the utility company that they are transparent when providing information, attitude towards and knowledge on DLC programmes. Utilities will need to carefully address these issues in the programme design to ensure a widespread acceptance.

Latent class choice model with a flexible class membership component: A mixture model approach

This study presents a Latent Class Choice Model (LCCM) with a flexible class membership component. Specifically, it formulates the latent classes using Gaussian-Bernoulli mixture models and investigates the impact of such formulation on the representation of heterogeneity in the choice process, goodness-of-fit measures and out-of-sample prediction accuracy of the choice models. Mixture models are model-based clustering techniques that have been widely used in areas such as machine learning, data mining and pattern recognition for clustering and classification problems. An Expectation-Maximization (EM) algorithm is derived for the estimation of the proposed model. Using two different case studies on travel mode choice behavior, the proposed model is compared to traditional discrete choice models on the basis of parameter estimates’ signs, values of time, statistical goodness-of-fit measures, and cross-validation tests. Results show that mixture models improve the overall performance of latent class choice models by providing better out-of-sample predication accuracy in addition to better representations of heterogeneity without weakening the behavioral and economic interpretability of the choice models.

Air Passengers’ Preferences for Aviation Voluntary Carbon Offsetting: A Co-benefits Perspective

Co-benefits are positive outcomes from voluntary carbon offsetting (VCO) programs beyond simple reduction in carbon emissions, which include biodiversity, air quality, economic, health, and educational benefits. Given the rates of aviation VCOs remain at less than 10%, this study investigated air passengers’ preferences for co-benefits as well as certification, location, and cost of VCO programs. Using discrete choice modeling, this study shows that aviation VCO programs with higher levels of co-benefits, particularly biodiversity and health benefits, are preferred by air passengers and confirms a preference for domestically based and certified VCO programs. The latent class choice model identified three classes with different preferences for VCO program attributes and demographic characteristics. The results of this study contribute to the knowledge of VCO co-benefits and imply that airlines should take note of this preference for biodiversity and health co-benefits when designing VCO programs and differentiate between market segments to increase the uptake of VCOs.

The potential impact of media commentary and social influence on consumer preferences for driverless cars

The emerging technology of autonomous vehicles (AVs) is envisaged to substantially alter transport systems and travel behaviour. However, this impact is conditional on the level and rate of AV market penetration. This study examines consumers’ consideration to adopt AVs, with a particular focus on the impact of informational cues from different communication channels. Data for our analysis is collected through an online survey conducted in December 2019, and our sample comprises 862 residents from Sydney, Australia. The survey asks respondents about their social networks, and their use of different mass and social media platforms. Respondents are subsequently presented a discrete choice experiment where we systematically vary informational cues about AVs from each of these communication channels, and respondents are asked to indicate their willingness to consider AVs under each scenario. This data is used to develop a latent class choice model of AV consideration as a function of the informational cues received from different communication channels. Our model also considers the level of trust that respondents associate with their social contacts when modelling the effect of word-of-mouth from these social contacts. The results reveal eleven segments in the sample that differ from one another in terms of sociodemographic attributes, evaluation of social contact trust, and the relative importance they attach to different communication channels. The findings show social media sentiment has the highest effect on AV consideration for 90 per cent of the sample. Also 3.8 per cent of the sample never consider adopting AVs, regardless of the informational cues.

Computer Says I Don\u2019t Know: An Empirical Approach to Capture Moral Uncertainty in Artificial Intelligence

As AI Systems become increasingly autonomous, they are expected to engage in decision-making processes that have moral implications. In this research we integrate theoretical and empirical lines of thought to address the matters of moral reasoning and moral uncertainty in AI Systems. We reconceptualize the metanormative framework for decision-making under moral uncertainty and we operationalize it through a latent class choice model. The core idea being that moral heterogeneity in society can be codified in terms of a small number of classes with distinct moral preferences and that this codification can be used to express moral uncertainty of an AI. Choice analysis allows for the identification of classes and their moral preferences based on observed choice data. Our reformulation of the metanormative framework is theory-rooted and practical in the sense that it avoids runtime issues in real time applications. To illustrate our approach we conceptualize a society in which AI Systems are in charge of making policy choices. While one of the systems uses a baseline morally certain model, the other uses a morally uncertain model. We highlight cases in which the AI Systems disagree about the policy to be chosen, thus illustrating the need to capture moral uncertainty in AI systems.

Investigating heterogeneity in preferences for Mobility-as-a-Service plans through a latent class choice model

The past decade has seen the introduction and widespread availability of a number of new mobility services. These have created a transport environment that is complex to navigate for passengers. The Mobility as a Service (MaaS) concept aims to provide a solution, by offering a single digital interface through which users can plan journeys, pay for and access a variety of transport modes. MaaS can also provide users with various products, including pay-per-use access to transport modes as well as MaaS packages. The latter are bundled mobility services that combine a variety of transport modes and are offered to customers in a one-stop-shop manner. The objective of this paper is to examine individual preferences for MaaS packages, specifically addressing the question of preference heterogeneity. In doing so, a Latent Class Choice Model (LCCM) is developed, allowing us to reveal variations in individuals’ preferences. The LCCM is estimated using data from a MaaS-related market research carried out in Greater Manchester. The results imply significant heterogeneity with regards to preferences. Three latent classes emerged through the analysis, all with different MaaS package preferences and individual characteristics. Age, gender, income, education and current travel behaviour all play an important role in determining an individual’s propensity to purchase MaaS packages. The results can provide valuable insights into the types of people that should and should not be initially targeted with MaaS packages to maximise uptake.

Disentangling the behavioural side of the first and last mile problem: the role of modality style and the built environment

The first and last mile (FLM) problem, namely the poor connection between trip origins or destination and public transport stations, is a significant obstacle to sustainable transportation as it is likely to encourage the use of cars for FLM travel, if not for the entire trip. This study examines the role of modality style and built environment in FLM mode choice behaviour, in order to identify the key features that might invoke a travel mode shift from cars to more sustainable travel options for both mandatory and discretionary trips. More specifically, this study draws on disaggregate data from the South East Queensland household travel survey and presents a latent class choice model to unravel modality style groups. Results reveal two distinct individual-level modality style groups: (1) driving and walking oriented; (2) multimodal travellers. Individuals in the second modality style group were found to be relatively inelastic to FLM travel time for mandatory trips, while individuals in the first group were largely unaffected by built environment characteristics and highly habitual in their mode choice behaviour for both mandatory and discretionary trips. Home residence environments with high road intersection density and public transport accessibility, and home residence environments with diverse land use mix, respectively encourage individuals within the second modality style to walk for mandatory trips, and discretionary trips. To this end, when place-based policies seek to change certain built environment features, individuals in the second modality style are more likely to shift their preference from cars to more sustainable modes. Finally, our findings have practical planning implications in targeting mode shift through highlighting the importance of considering the intersection of individual modality style in a given locale and mode choice behaviour. More specifically, our findings advocate for place-based policies that seek to target particular locales with the certain modality style deemed to be more predisposed to adopting a mode shift.

The Influence of Alcohol Warning Labels on Consumers’ Choices of Wine and Beer

This study aims to analyse the influence of alternative formats of health warnings on French and Italian Millennial consumers’ choices of beer and wine. Two Discrete Choice Experiments were built for wine and beer and two Latent Class choice models were applied in order to verify the existence of different consumer profiles. Results show that young consumers’ choices for wine and beer are influenced by framing, design and visibility of warnings. In both countries, the acceptance of warnings is higher for beer than for wine and in both cases consumers show higher utility for a logo on the front label: on the neck with a neutral message in the case of beer; on the front, without a message for wine. Latent Class choice models highlight the existence of different consumers’ groups with different levels of warning influencing their choices. In order to apply policies conducting to health benefits, our results suggest the need to focus on young individuals to communicate the risks of alcohol abuse through targeted messages and, more generally, to make them aware of the potential negative effects of excessive consumption of both wine and beer.

How habit moderates the commute mode decision process: integration of the theory of planned behavior and latent class choice model

This study attempts to develop a comprehensive framework by integrating the theory of planned behavior (TPB) and latent class choice model, with aim to understanding how mode-use habits moderate the process underlying commute mode choice. By designating habit as the covariate in the class membership model, three segments with unique mode-use habit style are obtained. First, heterogeneity in the effects of socio-demographic variables and TPB-related cognitive factors on commute mode choice across segments are empirically confirmed. Second, by directly including the whole TPB framework into the choice model, the decision-making mechanism underlying commute mode choice is explicitly reflected, which significantly varies with respect to specific mode choice in each segment. Either “a habitual and automatic behavior” or “a deliberate and rational decision” is finally determined. This study provides an empirical support to the moderating role of mode-use habit in the commute mode choice process, with a particular focus on its moderating role in the effect of TPB-related factors. The findings suggest that strategies to manage transport modes ownership and usage must be targeted towards specific population groups in order to gain effectiveness.

Quantifying preference heterogeneity in transit service desired quality using a latent class choice model

Abstract This study aims at quantifying preference heterogeneity in transit service desired quality to better-informing service quality improvements. The analysis is performed using a validated dataset elicited from 906 respondents through an online survey. An unlabelled Stated Preference (SP) experiment was utilized in a Latent class Choice Model (LCM), and an Error Components interaction model. The results of the EC interaction model revealed preference heterogeneity due to differences in respondents’ socioeconomic and behavioural characteristics. While the results of the LCM untapped vital information that has not been reported previously in the transit service quality literature. Unlike the traditional user type classification, our study classifies respondents into three segments: Direct Trip Enthusiastic (DTE), Cost-Sensitive (CS), and Real-time Information Supporter (RIS). Each segment exhibits different preferences for transit service attributes, and their willingness to pay for service improvements is distinctly different. Further, the LCM indicates that the heterogeneity of users’ preferences is not explicit in their usage pattern nor accessibility to different travel modes; instead, it is a bundle of various parameters.

Modelling heterogeneity in behavioral response to peak-avoidance policy utilizing naturalistic data of Beijing subway travelers

Studies of travelers’ response behavior to transportation demand management is receiving substantial attention among researchers and transport operators in recent years. While previous studies in this area have generally assumed that the sensitivity of travelers to different factors is homogeneous and relies on survey responses, which may be prone to self-reporting errors and/or subject to behavioral incongruence. Relying on naturalistic data, this paper aims to investigate the behavioral response to pre-peak discount pricing strategy in the context of the Beijing subway with a special focus on the heterogeneity among the travelers. Anonymous smart card data from 5946 travelers before and after the introduction of a peak avoidance policy in Beijing are used to construct a latent class choice model to capture the sensitivity to different factors and the associated taste heterogeneity of travelers. Given the passive nature of the data, the model can offer more realistic outputs. The results indicate that there is substantial heterogeneity in travelers’ responses to the peak avoidance policy, and that they can be probabilistically allocated to four latent classes. For all classes of travelers, the decision to shift their departure to off-peak is affected by the monetary saving, the required change in departure time and the frequency of travel, but in different magnitudes. In particular, only two classes of travelers (who exhibit lower standard-deviation in pre-intervention departure time) show significant sensitivity to price changes indicating that the discount policies are more likely to be effective for these groups. The rest of travelers are largely price insensitive – warranting the need for non-monetary incentives as opposed to fare discounts. To the best of our knowledge, this study is the first to innovatively apply the LCC framework to analyze travelers’ heterogeneous behavior using large-scale smart card data without socio-demographic information. The findings can provide guidance to the subway authority in devising differential peak avoidance policies targeted for different groups of users, which are likely to be more effective than the current ‘one size fits all’ approach.

Determining the Market Uptake of Demand Responsive Transport Enabled Public Transport Service

Demand responsive transport (DRT) alternatives offer improved mobility to travellers through station-to-destination or door-to-transit operations. In particular, door-to-transit DRT service acts as a feeder to major public transport hubs, making public transport more accessible and attractive to travellers. This work aims to study the mode choice behaviour of travellers between their current modes and a new service, which is a combination of DRT and public transport. The study is conducted in the Northern Beaches area of Sydney, Australia where DRT is expected to serve as a feeder to the newly introduced express bus service called B-Line. A stated preference (SP) experiment is designed where multiple-choice scenarios involving two modes, status quo (SQ) and the new service (combined DRT and public transit), are presented to the participants. The survey uses trip specific information obtained from Google API to form the attributes for the new service. The collected data are analysed using a latent class choice model (LCCM), which segments the observed sample into distinct groups where each group has its own taste and preferences towards the new service option. Results from the study reveal that one of the identified user segments shows 96 percent uptake towards the new service option, while the other user segment shows an uptake of 44 percent. Results also show that individuals making work trips are more likely to opt for the new service. Findings from this study can provide information to urban planners regarding the market uptake of DRT services. Furthermore, the findings can also help planners in implementing segment specific policies aimed at further improving uptake towards DRT along with public transport.

Identifying user classes for shared and automated mobility services

New forms of shared mobility such as free-floating car-sharing services and shared automated vehicles have the potential to change urban travel behaviour. In this paper, we identify potential user classes for these new modes. For this, a stated choice experiment on mode choice among a sample of the Dutch urban population has been conducted, which features free-floating car-sharing and shared automated vehicles next to private vehicles, bus, and taxi. The experimental design allows disentangling the effects of vehicle ownership, vehicle sharing and vehicle automation on the perceived utility of these modes. Further contributions lie in the identification of user classes for shared and automated mobility services and their potential migration from their current modes to the these services. Latent class choice models were estimated to capture the heterogeneity in these preferences among the respondents. The most explanatory mode choice model is obtained by estimating a 3-class nested logit model capturing the impact of vehicle ownership. The results show that higher educated and more time-sensitive respondents are more inclined than others to favour the (automated) car-sharing options. By simulating a scenario that directly compares car with free-floating car-sharing and taxi with shared automated vehicles, a migration analysis has been performed. This analysis shows that the preferences towards shared automated vehicles and free-floating car-sharing is highest for those currently combining car and public transport for their commute. Commuters using the car showed a high preference towards free-floating car-sharing, in particular as for the latter no parking fees are issued. Respondents currently commuting by public transport showed the lowest preference for the new modes.