In travel behaviour modelling, latent class models are used to represent underlying discrete groupings of behavioural preferences. The paper presents a latent class extension of a dynamic discrete choice model (DDCM) and applies the model to the problem of activity demand generation and scheduling. The DDCM is a recursive multinomial logit model where agents make sequential decisions in time, maximizing the expected future utility of their decisions in a random utility maximization framework. It generates activities and their associated travel within a full day schedule, endogenously respecting agents' inherent time-space constraints. The latent class DDCM builds on the base model by representing heterogeneous lifestyle preferences. A specification of the model is estimated on a Stockholm travel survey and uses age, income level, gender, car ownership and presence of children in the household as classifying variables. The models result in classes which primarily represent modality styles, finding car-, transit- and bike-primary behavioural groups as well as a multimodal group, each linked with different socio-demographic characteristics. The models improve over non-latent class reference models and provide insight into the structure of heterogeneity in travel behaviour preferences in Stockholm.

We present the use of duration-dependent activity utility within the dynamic scheduling model Scaper, which simulates individuals' full-day activity and travel schedules. In Scaper, agents make sequential choices in time which maximize expected future utility and respect time-space constraints. Using Swedish travel survey data, we estimate a new version of the model including piecewise linear utility functions for marginal activity duration by activity purpose. Our model reveals a strong duration dependence for work, leisure, and visit activities with differing functional shapes for each purpose. In simulation, the duration-dependent model better reproduces observed distributions of activity duration and performs as well across other metrics as the model without duration dependence. We illustrate the potential policy applications of the model using a scenario of shortened work days. The duration-dependent model offers useful predictions for the effects of the scenario on commute timing, nonwork activities, time spent at home, and trip chaining.

This paper applies recursive logit (RL) to model activity-trip chaining behaviour. We present a comparison between two approaches to applying the RL model in this context. In the first ‘sequential’ approach, agents form a trip chain by making a sequence of joint choices of activity location (i.e. trip destination) and travel mode, ending the chain by choosing to return home. The second ‘dynamic’ approach adds a time variable. Its agents form a full-day activity/travel schedule by making a sequence of choices either to continue the current activity for a fixed timestep or make a joint choice of new activity location and travel mode. We estimate parameters for both models using data from a Stockholm travel survey and validate model simulations against observed data. The models reproduce patterns of observed behaviour beyond their estimated parameters, including different types of trip chains and the spatial distribution of activities. While the dynamic model is advantageous in its ability to predict agent schedules, reflect time-varying travel conditions and endogenously represent space–time constraints, it does not surpass the simpler sequential model on mutual areas of trip chaining behaviour. We conclude that the RL model is well-suited to model trip chaining behaviour, and that the simpler sequential approach may be appropriate for many modelling purposes.

Travel behaviour models are key decision support tools for transport planning that point the way to more sustainable and equitable transport systems and land use. Activity generation and scheduling models are a type of travel behaviour model which consider both travel and activity participation in a temporal context, reflecting the importance of time to understanding daily travel decisions. This thesis dynamically models individuals’ activity schedules, with emphasis on reflecting the inherent complexity of travel behaviour and the influences and constraints of space-time contexts.

The thesis applies and extends a dynamic discrete choice model for activity generation and scheduling. Discrete choice models have a well-developed theory encompassing behaviourally-consistent models with estimable, meaningful parameters and microeconomic welfare metrics. Dynamic models endogenously represent time, allowing them to represent interdependencies across time and making their outputs time dependent. The model used in the thesis, Scaper, started from the work of Karlström (2005). Its agents make chronological joint choices of activity purpose, location and timing. They are forward looking, making decisions which maximize their expected future utility. The model generates full-day activity schedules consistent with an agent's space-time constraints.

Within activity generation and scheduling, two papers in the thesis focus on representing the inherent complexity in people's daily travel behaviour patterns. Paper 3 establishes that trip chaining behaviour can be seen as an emergent property of a dynamic decision problem and shows how the resulting dynamic model accurately predicts several facets of trip chaining behaviour including activity purposes, locations, timings and durations. Paper 4 addresses the heterogeneity of individual behaviour using a latent class approach, demonstrating the value of the method and finding evidence for several distinct travel behaviour lifestyles primarily but not solely differentiated by travel mode preferences.

In recognition of the central role of time in activity scheduling, another two papers aim to improve the allocation and valuation of time in the dynamic scheduling context. Paper 1 concentrates on time allocation, improving the dynamic model's predictions of activity duration by making the utility rate of participating in an activity dependent on its elapsed duration. Paper 5 derives two measures of the value of travel time savings in the dynamic context, one dependent on time of day and the other reflecting the expected value of time savings to an individual at the start of the day.

Finally, the thesis focuses on understanding how people's spatial context relates to their travel behaviour. Paper 2 evaluates urban development patterns of centralization vs. dispersion with respect to travel behaviour. The paper uses a work location model developed by Naqavi et al. (2023) paired with Scaper to establish agents' spatial contexts across three scenarios, and analyses different aspects of their predicted travel behaviour. It reveals a tension between planning goals, where centralization leads to more sustainable mode choices while dispersion may help reduce residential segregation.

Modeller för resbeteenden är viktiga beslutsstödsverktyg i transport-planeringen, som visar vägen där transportsystem och markanvändning blir mer hållbara och rättvisa. Modeller för aktivitetsgenerering och schemaläggning är en typ av resbeteendemodell som beaktar både resor och deltagande i aktiviteter i ett tidsmässigt skeende, vilket speglar dess betydelse för att förstå dagliga resebeslut. Denna avhandling modellerar dynamiskt individers aktivitetsplaner, med fokus på att återspegla den inneboende komplexiteten i resbeteenden och hur de rumsliga och tidsmässiga sammanhangen påverkar och begränsar resbeteendena.

I avhandlingen tillämpas och vidareutvecklas en dynamisk diskret valmodell för aktivitetsgenerering och schemaläggning. Diskreta valmodeller har en väletablerad teori som omfattar beteendemässigt konsistenta modeller med estimerbara och meningsfulla parametrar samt mikroekonomiska välfärdsmått. Dynamiska modeller representerar tid endogent, vilket gör att de kan fånga tidsberoende samband och producera tidsberoende resultat. Modellen som används i denna avhandling, Scaper, bygger på Karlström (2005). Modellens agenter gör kronologiska sammanhängande val av aktivitetsändamål, plats och tidpunkt. De är framåtblickande och fattar beslut som maximerar deras förväntade framtida nytta. Modellen genererar aktivitetsplaner för hela dagen som är förenliga med agentens rumsliga och tidsmässiga begränsningar. 

Inom aktivitetsgenerering och schemaläggning fokuserar två av avhandlingens artiklar på att representera den inneboende komplexiteten i människors dagliga resbeteenden. Artikel tre visar att reskedjor kan ses som en egenskap som uppstår i ett dynamiskt beslutsproblem och visar hur den resulterande dynamiska modellen väl predikterar flera aspekter av beteenden för länkning av resor, inklusive ändamål, platser, tidpunkter och varaktigheter för aktiviteter. Artikel fyra tar upp heterogeniteten i individuella beteenden genom att använda en metod för latent klasser, som visar värdet av metoden och ger evidens för flera distinkta livsstilar för resbeteenden, främst men inte enbart, differentierade efter preferens för transportmedel.

Tiden spelar en central roll i individernas schemaläggning av aktiviteter, därmed syftar två av artiklarna till att förbättra allokering och värdering av tid i dynamisk schemaläggning. Artikel ett fokuserar på tidsallokering och förbättrar den dynamiska modellens prediktion av aktiviteters varaktighet, genom att justera nyttan av deltagande i en aktivitet baserat på den förflutna varaktigheten. Artikel fem härleder två mått på värdet av tidsbesparingar i dynamiska kontexter, där det ena är beroende av tidpunkt under dagen och det andra speglar det förväntade värdet av tidsbesparingar för en individ i början av dagen. 

Slutligen fokuserar avhandlingen på att förstå hur människors rumsliga kontext relaterar till deras resbeteenden. Artikel två utvärderar urbana utvecklingsmönster av centralisering kontra spridning i förhållande till resbeteenden. Artikeln använder en modell för arbetsplatslokalisering utvecklad av Naqavi et al. (2023) tillsammans med Scaper för att fastställa individers rumsliga kontexter i tre scenarier, och analyserar olika aspekter av deras predikterade resbeteenden. Artikeln visar en motsättning mellan planeringsmålen, där centralisering leder till mer hållbara val av transportmedel medan spridning kan bidra till att minska den bostadssegregationen.

La modélisation des déplacements est devenu un outils clés d’aide à la décision dans le cadre de la planification des transports, permettant notamment, d’orienter l’aménagement du territoire et des systèmes de transports vers une trajectoire plus durable et équitable. Les modèles de génération et de planification des activités sont un type de modèle qui prennent à la fois en compte les déplacements et la participation aux activités de la vie quotidienne dans un cadre temporel, reflétant ainsi l’importance de la temporalité dans les choix quotidiens des voyageurs. Cette thèse modélise de manière dynamique l’organisation temporelle des activités des individus, en réfléchissant tout particulièrement à la complexité inhérente aux déplacements individuels ainsi que l’influence et les contraintes imposés par les contextes spatio-temporels. 

La thèse applique et développe un modèle dynamique des choix discrets pour la génération et l’organisation d’activités. Les modèles des choix discrets reposent sur une théorie déjà bien développée qui comprend à la fois des modèles de déplacements constants avec des paramètres estimables et significatifs ainsi que des indicateurs de bien-être microéconomiques. Les modèles dynamiques représentent le temps de manière endogène, ce qui permet de rendre compte des interdépendances à travers le temps, et de rendre leurs résultats dépendants de la temporalité. Le modèle « Scaper » mobilisé dans la thèse a été élaboré à partir des travaux de Karlström (2005). Dans ces travaux, les agents font des choix chronologiques conjoints, d’objectif, de lieu, et de temps de leurs activités. Ils sont tournés vers l'avenir, prenant des décisions qui maximisent leur utilité future espérée. Le modèle génère des horaires d'activités sur une journée complète conformes aux contraintes spatio-temporelles d'un agent.

Concernant la génération et la planification d'activités, deux articles de la thèse s’évertuent à représenter la complexité inhérente aux déplacements quotidiens des individus. L'article 3 établit que le chaînage des déplacements peut être vu comme une propriété émergente d'un problème de décision dynamique et montre comment le modèle dynamique qui en résulte prédit avec précision plusieurs facettes de ces chaînages, y compris l’objectif des activités, leurs localisations, horaires et durées. L'article 4 aborde l'hétérogénéité du comportement individuel en utilisant une approche de classe latente.  Il démontre à la fois l’apport de la méthode et l’existence de modes de déplacements distincts différenciés principalement, mais pas entièrement, par les préférences modales. 

Reconnaissant le rôle central du temps dans la planification d'activités, deux autres articles visent à améliorer l'allocation et l'évaluation du temps dans le contexte de la planification dynamique. L'article 1 se concentre sur l'allocation du temps, et améliore la manière dont le modèle dynamique prévoit la durée des activités en faisant dépendre le taux d'utilité de la participation à une activité à sa durée écoulée. L'article 5 dérive deux mesures de la valeur des gains de temps dans le contexte dynamique, l'une dépendant de l'heure de la journée et l'autre reflétant la valeur attendue du gain de temps pour un individu au début de sa journée.

Enfin, la thèse s’attache à comprendre la manière dont le contexte spatial des individus joue un rôle dans les comportements de mobilité. L'article 2 évalue l’effet de la concentration et de l’étalement urbain sur les déplacements. Il utilise un modèle de localisation de l’emploi développé par Naqavi et al. (2023) associé à Scaper pour établir les contextes spatiaux des agents à travers trois scénarios et analyse différents aspects de leur déplacements prédits. Il révèle une tension entre les objectifs de planification, où la centralisation conduit à des choix de mode plus durables tandis l’étalement peut aider à réduire la ségrégation résidentielle.

The City of Stockholm is conducting a scenario planning exercise to explore where potential future office development should be planned: closer to the city centre as in the status quo, in peripheral hubs on the outskirts of the city, or dispersed throughout multiple neighbourhoods. To support this exercise, this paper models these three scenarios using a nested work location and dynamic activity-based scheduling model. Our model predicts that high-income individuals have the highest consumer welfare benefits and are over-represented as workers in all scenarios. Developing more central office space will likely reinforce existing geographical patterns of income inequality in Stockholm; developing peripheral or dispersed office space, especially in the south of the city, will challenge these patterns. However, the model also illustrates a tension between the goals of equity and the environment. By taking advantage of existing transit infrastructure and congestion patterns, more central office development will result in lower vehicle kilometers travelled and lower car mode share for commuting than more peripheral or dispersed development.

This paper improves a dynamic scheduling model, Scaper, by including dependence between agents' mode choice and activity purpose. Observed differences in mode share across activity purposes are likely caused both by factors endogenous to the model such as differing levels of service by time of day or geography, and by factors which cannot be represented endogenously with the data available, such as perceived travel needs. The updated model introduces joint purpose- and mode-dependent parameters to account for these exogenous factors. The model improves on a reference model without joint parameters on mode share predictions of activities of different purposes, providing advantages for use in activity purpose-sensitive policy evaluation.

This paper reports the development of a microsimulation model-building platform and the implementation, validation, and deployment of an integrated transportation and land use model using the platform. The modeling platform provides a new approach to microsimulation by providing modular algorithms with which models can be configured without additional coding, enabling the rapid development, testing, and deployment of models. The integrated transportation, land use, and energy (iTLE) model is an operational microsimulation model for Halifax, Canada, that combines long-term household decisions with short-term travel choices. The model’s longer-term decision simulator simulates life-stage transitions, residential mobility, and vehicle transactions. The residential mobility module uses a logsum measure of modal accessibility to capture how house-holds’ travel options influence their location decisions. The model was run from 2006 to 2036 on a business-as-usual scenario of land use and transportation. Validation performed after 10 years against census data indicates that the model produces a well-calibrated simulated population. The model predicts rising population density and greater household incomes in the urban core and a rise in vehicle ownership. The disaggregate nature, integration of land use and travel behavior, and life-course perspective of the modeling framework make it valuable as a decision-support tool for testing alternate scenarios against the baseline. The ease of development afforded by the modeling platform will facilitate future research on new mobility options, ensuring that the model can easily be adapted to anticipated transformational changes in the transportation system. 

We present an approach to simulating individuals' daily trip and activity chains which treats these as the emergent behaviour of a dynamic decision problem. In our framework, agents in a Markov Decision Process try to maximize the total reward of their path through the state space. As agents observe the immediate reward of transitions and the expected utility of the resulting states, their actions are modelled using a recursive logit model. We present three recursive logit model specifications for trip chaining: destination choice only, mode and destination choice, and a fully dynamic model which adds activity scheduling. We estimate parameters for these models using data from a Stockholm travel survey and validate the models against observed data in simulations. The models produce trip chains that reproduce observed behaviour including the distribution of activities per tour and trip chain patterns. The dynamic model reproduces observed trip chain timing and produces reasonable variability in behaviour for agents with different home and work locations. The dynamic model is advantageous in its ability to predict dynamic trip chaining behaviour, reflect time-varying travel conditions and endogenously represent space-time constraints.