Search

In response to insufficient understanding of the determinants of change in e-shopping behaviors during the COVID-19 pandemic in developing countries, this paper used the data from 355 respondents, collected in Hanoi during the social distancing period (April 2020), to explore the factors associated with shopping online more frequently (i.e., representing the growth of e-shopping) for five product types (food, medical products, clothing, electronics, and books) in Hanoi, Vietnam. The results showed that nearly 80% of the respondents engaged in e-shopping more frequently than they did before the outbreak of COVID-19. As regards shopping online more frequently in general (i.e., for at least one product type), females were more likely to do so. In-store shopping enjoyment and a decrease in income were a facilitator and a deterrent, respectively. Regarding specific product types, completely working from home had a positive association with more frequent e-purchasing for electronics. Fear of disease encouraged higher frequencies of e-shopping for food and medical products. Notably, the shortage of physical supply was not a determinant of buying any product type online more frequently. As for the implications of our findings, supporting and encouraging low-income shoppers, older persons, and females to engage in e-shopping is necessary to limit the detrimental effects of the pandemic on their lives. The growth of internet purchasing expresses a need to manage the development of urban delivery services, to limit the uncontrolled proliferation of motorcycles. E-shopping requires delivery to complete the online-to-offline process, therefore, protecting the health of delivery drivers to ensure the safety of the whole online shopping process would be necessary.

Recently, there has been an increasing interest in targeting the safety of bus operations worldwide, however, little is known about the determinants of the bus crash severity in developing countries. By estimating an ordered logit model using the bus-involved collision data in Hanoi (Vietnam), spanning the period from 2015 to 2019, this study investigates various factors associated with the crash severity. The results reveal that the severity risk increases for (1) large buses, (2) raining conditions, (3) evening or night, (4) sparse traffic, (5) non-urban areas, (6) roads with at least three lanes, (7) curved roads, (8) two-way roads without a physical barrier, (9) head-on collision, and (10) pedestrian-related crashes. Aside from confirming the crucial roles of a wide range of factors, this research has examined the effects of two determinants (traffic density and crash area) that have not been considered for the cases of developing countries previously. Based on the findings on the impacts of factors, a series of policy recommendations regarding improving road conditions in non-urban areas, promoting walking infrastructure, reminders of high-risk situations for drivers, safety notes when improving bus service quality, and recording bus-related crashes are proposed.

Walking plays an essential role in both daily mobility and human health. Unfortunately, measuring walking behavior in self-reported surveys is usually subject to lack of accuracy due to the limit of human memory, the habit of not only rounding travel time but also neglecting short and unimportant trip legs. To cope with such a problem, the use of the Global Positioning System (GPS) is promising. This study aimed at developing an inference model to derive travel modes from 406-day GPS data of 90 respondents living in Lyon city and its suburbs (France). Based on mode detection results, aspects of walking (i.e., purposes, length, duration, speed, the relation between walking and home, the relation between walking and other modes) were estimated. For detecting modes well, particularly walking, a fuzzy logic-based algorithm using average speed, the 95th percentile of speed, and the 95th percentile of acceleration were created. Rules of signal loss were proposed to identify metro segments, while respondents’ reports on the frequency of cycling were utilized to address the confusion between walking and cycling better. As for findings, 951 walking stages were found. Because all of the walking patterns, estimated from prediction results of travel modes, were compatible with previous studies’ findings, the mode detection method adopted in this study has performed satisfactorily, particularly concerning imputing walking. The proposed method may enable better use of GPS to conduct further analyses of walking behaviors thanks to deriving walking segments adequately from spatio-temporal data.

The Global Positioning System (GPS) has motivated rapid advances in mobility data collection. A massive amount of spatio-temporal information has made it possible to know where a person was and when, but not how and why (s)he travelled, creating the need for inference models. Compared with mode detection, purpose imputation has been insufficiently studied. However, the relative lack of attention to purpose identification does not mean that this field has not emerged. For this paper, which is the first review dedicated to inferring trip purposes from GPS data, 1,162 non-duplicate papers from four databases (Scopus, Web of Science, ScienceDirect and TRID) were screened, and a corpus of 25 publications was selected for examination. Based on these papers, the purpose imputation problem is defined in the contexts of the evolution of GPS-based travel surveys and two research domains, Transportation Science (TS) and Human Geography (HG). Subsequently, three steps of the purpose detection process, namely (1) trip end detection, (2) input feature selection and (3) main algorithms and validation, are surveyed. During these procedures, the differences between studies in TS and those in HG are highlighted. Finally, unresolved issues related to data and feature selection, algorithms and assessment are discussed substantially to provide potential research directions. This review may be an informative reference for those newly accessing the GPS-based purpose imputation field and/or intending to develop solutions to this problem.

Transportation mode detection is one of the major challenges in GPS-based travel surveys. This study presents attempts to impute modes from data collected by smartphone in Hanoi (Vietnam), where the dominant mode is the motorcycle. The inclusion of the motorcycle mode and an imbalance in the modal share of the Hanoi data resulted in the ineffective use of supervised-learning models to detect all modes simultaneously. To gain a high level of accuracy and reasonable interpretability, a hierarchical process was developed. At first, walk, bicycle, and motorized modes were identified by a fuzzy logic-based algorithm. Subsequently, based on the distribution of bus stops and the operation of buses in practice, rules using the average distance between stops that a vehicle passed slowly or stopped at were introduced to detect bus segments. Finally, a random forest model was built to distinguish the modes of motorcycle and car. The proposed hierarchical process achieved an accuracy level of 89.1%. The bus detection, which required coordinates of bus stops solely, achieved a recall of 87.2%. The mode of motorcycle was the main source of misclassification. Including the motorcycle mode has contributed to the diversity of the mode detection field, which has previously focused on walk, bicycle, car, bus/tram, and train. The hierarchy was developed and validated by using a dataset that did not include travel by metro or train and would be biased toward persons working and studying at a university. These limitations emphasize the need to test the process on a more diverse sample with more travel options.

This paper presents a method for enhancing purpose imputation from global positioning system data without using geographic information system data via relevant feature selection from six groups: (1) activity time; (2) user characteristics; (3) predicted travel modes; (4) actual travel modes; (5) estimated home location; and (6) estimated location of the most frequently visited non-home place (MFVP). Two datasets were collected in 2019 using TRavelVU, a smartphone application. The first one (the Hanoi dataset) comprised 652 days’ worth of data collected from 63 users in Hanoi, Vietnam, whereas the second one (the Donate dataset) comprised 932 days’ worth of information collected from 65 individuals in Denmark, Sweden, and Norway. The hyperparameters of the random forest models were tuned carefully in accordance with selected features, thereby facilitating a thorough evaluation of the improvement in prediction models. The findings of this study revealed that the addition of either actual or predicted modes resulted in improved imputation performance, albeit the former exhibited a stronger positive effect. This demonstrated the potential benefits of integrating mode detection and purpose identification into a continuous process. The newly adopted MFVP feature contributed to enhanced prediction results (around 2%). The proposed purpose-imputation models, which benefited from all features, demonstrated accuracies of the order of 75% and 85% for the Hanoi and Donate datasets, respectively. The imputation of home and work/education activities demonstrated high success, whereas reasonable prediction results with nearly all F-score levels ranging between 50% and 83% were observed for pick-up/drop-off, shopping/eating, visit/leisure, and business activities.

During this workshop on "representativeness in surveys : challenges and solutions" for mobility surveys, we discussed various issues of representativeness in travel surveys. Issues such as lack of coverage of sampling frames, nonresponse mechanism and measurement error bias were investigated along with models that can help to reduce the impacts of these issues and preserve data quality for transport analysis. According to the participants of the workshop, the best way to avoid problems with representativeness in surveys is to tackle the problem upstream (good sampling frame, follow-up with respondents, incentives, response facilitators, etc.). But still at the end of the day we will need to reweigh the respondent sample to account for differential probabilities of selection among subgroups; effects arising from nonresponse; inadequacies in sample frame, etc. and bring the respondent sample data up to the dimension of the study population.

This paper investigates trends in the travel behaviour of young adults in Germany, France, Great Britain, Japan, Norway, and the USA over the past few decades with a focus on car availability and car travel. The trend analysis relies on micro-data from over 20 National Travel Surveys from the study countries dating back to the mid-1970s. The analysis of the survey data is supplemented by official statistics on licence holding. On this basis, this paper compiles a body of evidence for changes in mobility patterns among young adults in industrialized countries over the past few decades. The findings indicate that since the turn of the millennium, access to cars, measured in terms of drivers' licences and household car ownership, has decreased in most study countries—especially for men. Moreover, average daily car travel distance has decreased in most study countries, again especially for men. In France, Japan, and most significantly in the USA, the decrease in car travel has led to a reduction in total ever...

International audience; Ageing of the population, urban sprawl and car dependency will change travel patterns. The main objective of this paper is to give elements for a better understanding of the impact of changing demographics on the long term evolution of daily mobility using demographic-based models to forecast, for the elderly population, car-ownership, trip frequency, distance traveled, average trip distance. A second objective is to measure the impact of the long term tendencies observed on the appearance of new needs of travel demand such as a rapid increase of demand-responsive transport. The paper compares two agglomerations, both in a strong ageing process, but in quite different sociocultural contexts: a large European metropolis: Paris, and a medium sized north-american city: Montreal. Many common conclusions derived from the two different cases studies reinforce the possibility of generalizing the conclusions to other situations.

This paper discusses available household survey data on long-distance travel (LDT) and presents harmonized figures on LDT in Europe. First, there is a comparison of the results of different household travel surveys about LDT. The findings of this comparison have important implications for the methodology of surveying LDT: conventional mobility diary surveys are better than LDT surveys in capturing journeys up to 200 km. LDT surveys perform better in capturing travel only beyond 400 km. Second, the paper presents the first internationally comparable figures on LDT demand for Europe. These figures have been compiled by using different sources of information, avoiding the drawbacks of the different surveys. The results indicate that in northern and central Europe there is higher demand for LDT than in the south. Moreover, the results illustrate how policies and public-transport supply affect the modal split in LDT: in countries with a high-quality railroad supply, such as Switzerland, the railroad share is significantly higher than in other countries. Countries with a liberalized, interurban bus market, in contrast, have a significantly higher bus mode share.

Uncertainty on traffic forecasts may have an impact on reimbursement scheduling for investment, as well as for scenarios for operating costs. Even the best projections are based on models and assumptions, thus raising the question of their accuracy. Indeed, long term investments are risky and it is important to cope with uncertainty. This paper deals with the uncertainty on a long term projection with an Age-Cohort approach. The authors used the jackknife technique to estimate confidence intervals and observe that the demographic approach outlines the structural determinants for long term trends of mobility.

Nationwide Transport Surveys and Time‐Use Surveys both reflect the daily agendas and schedules of the reporting individuals and should therefore yield comparable indicators of travel behaviour; for instance: immobility rate (share of persons not leaving the home on any one day), daily travel time, and number of trips per day. These two surveys exist in three countries from the same time period: Belgium, France, Great Britain. The comparisons demonstrate that they tell parallel stories, but that the levels of the variables are significantly different with lower immobility rates and longer travel times reported in the Time‐Use Surveys. These surveys should therefore be integrated in the analysis of travel behaviour analysis as a crucial yardstick. In Europe, where Nationwide Travel Surveys are intermittent and not harmonized, the harmonised Time‐Use Surveys allow for crucial European‐wide comparisions across time and space.

Long term forecasting of mobility is a major challenge for urban transport planning. The demographic approach based on data from repetitive surveys makes it possible to get insight in the dynamics of behaviours for individuals belonging to different generations at various stages of their life cycle. The decomposition of temporal effects into an effect of age and an effect of generation makes it possible to draw the sample profile during the life cycle: it is the fundamental concept of the ?age-cohort? model to predict urban mobility in long term. The application of the model relates to the agglomeration of Lille, where we have three mobility surveys at 11 years intervals.

Resume Le projet Pari 21 n'a pas pour ambition de justifier un changement radical de systeme de transport — pratiquement tout a ete dit sur les avantages et inconvenients de l'automobile dans la ville. En revanche, le projet va simuler et tester, du point de vue de l'individu, la faisabilite d'un systeme de transport fonde principalement sur l'usage d'autres modes que le vehicule individuel dans la zone fortement peuplee de l'agglomeration parisienne (Paris et petite couronne). En construisant et evaluant des scenarios de transport, le projet vise a mesurer les marges de manceuvre dont on dispose pour reduire l'usage de l'automobile dans les zones urbanisees, en particulier celle qui resulte des baisses de vitesse de deplacement associees a ces scenarios. La methode repose sur une succession d'iterations d'un modele simulation combinant des affectations des deplacements a d'autres modes que l'automobile et des redefinitions de l'offre de transport. La demande est orientee vers les modes individuels (marche a pied, bicyclette), les lignes de transport en commun et des combinaisons des deux. Les iterations s'appuient sur I'Enquete globale de transport dIle-de-France de 1991–1992, qui fournit toutes les boucles de deplacement effectuees en automobile un jour donne par tons les habitants d'une zone (autrement dit toutes les sequences de transport entre l'instant ou la personne quitte son domicile et l'instant ou elle y revient).

Nonresponse results in a reduced sample size, but a more important concern of researchers is the possible impact of nonresponse bias. Bias is introduced when those who do not respond to the survey are systematically different from those who do respond on key variables of interest. The response mechanism is defined as “ignorable” when it can be modeled with the available characteristics (e.g., from the sample base). For example, if the nonresponse mechanism depends on the level of income, the respondent sample must be reweighted according to the distribution of income to correct the nonresponse errors. For the French National Travel Survey 2007–2008, the sample was drawn directly from the census and the list of new residences built since then. Therefore, there is considerable information about respondents and nonrespondents. The response mechanism is modeled first, and then calibration strategies are discussed. The nonresponse biases are quantified by using auxiliary information in different calibration exercises.

International audience; In this paper we challenge the prevailing practice of conducting one-off cross-sectional mobility surveys, making a case for change on the basis of usefulness and cost-effectiveness. We believe that urban areas over say, one million inhabitants, should collect mobility data on a continuous basis as part of their efforts to guarantee sustainable development. This would allow them to gain a proper understanding of the pressing environmental and transport-related issues of today's world, as well as of the effects of economic growth and price (especially for fuel) volatility. In our scope we include panel data (i.e. information from a smaller sample of respondents who are interviewed at different points in time, hopefully during several years), as this type of information is unique in its ability to help understanding behavioural changes and the impact of time-related effects, such as habit and inertia. We revise the state of affairs in different parts of the world, not limiting ourselves to urban data. Continuous surveys, on-going surveys, panel surveys.

Transport Survey Methods: Keeping Up With a Changing World, ISBN:978-1-84855-844-1, ISBN:978-1-84855-845-8

This chapter explains the methodology used to cope with the estimation of energy consumption in the new 2004 survey: the analysis made on previous surveys data to test this possibility and the different improvements made to the survey methodology to adapt the questionnaire to provide the distance calculation and the checking of the transport chain coherence and also to optimize the sampling.

International audience; The aim of the present study is to evaluate the possible extent of modal shifts from car use to 'alternative modes' (public transport, cycling, walking) without any change in individual patterns of activity. Its approach is based on a transfer procedure that allows the simulation of the maximal potential market for transport modes other than the private car. The method is based on repeated iterations of a simulation model that assigns journeys to transport modes other than the automobile based on a number of improved public transport scenarios. Demand is channelled towards individual modes (walking, cycling), public transport, and a combination of individual and public modes, based on their relative time and distance performance. The modal transfer procedure is applied to several transport supply scenarios, which provide a picture of what is possible in the sphere of modal split. Each simulation entails a potential transfer of private vehicle-km to each of the other modes. Even where different public transport scenarios are simulated, the transfer is evaluated for round trips in both the Paris and Lyon surveys. There is therefore no modification in the activity pattern of the people surveyed nor trips induced by improvements in transport supply. The aim is not to predict what would be the modal split in other circumstances, but the upper limit of the shifts. This paper presents our methodology and the principal results obtained through numerical simulations based on figures for the Paris and Lyon conurbations. This approach demonstrates that a policy focused on modal shifts has the potential to reduce car use, but that this potential is limited. Any aspiration to reduce car use further would mean changes in the patterns and location of activity.

Car Free City : Utopia ?. - Improvements of transport systems in big conurbation (higher speed and lower cost of individual mobility) have contributed to the reduction of urban areas density and have led to dominant use of the car and a dramatic reduction in walking and cycling, while the use of public transport has remained roughly constant. The car now highly dominates the other transport modes in many places including the Paris metropolitan areas. As a consequence of the car domination, a great part of the public opinion claims a reduction of the car use and the development of alternative forms of transport. The aim of this study is to determine whether those claims are consistent with the actual present speeds of each mode of transport. More precisely, our aim was to try to answer some questions : have car drivers a good appreciation of the car speed performance ? How many are car's drivers who could save time by using other modes of transport ? What can be expected from a drastic growth of the public transport supply and/or from a reduction of the car speed in order to reduce the car usage and consequently the car traffic flows. This paper presents our methodology and the major results obtained through numerical simulations based on Paris and Lyon conurbations figures., L'amélioration des systèmes de transport dans les agglomérations (vitesse accrue et baisse des coûts de transport) a contribué à la diminution de la densité dans les zones urbaines et a conduit à un usage dominant de la voiture et une très forte réduction de l'usage de la marche et du vélo, tandis que les transports en commun sont restés globalement stable. L'usage de la voiture est maintenant dominant face aux autres modes de transport y compris dans la métropole parisienne. Cette domination de la voiture conduit de plus en plus l'opinion publique à réclamer une réduction de son usage et le développement d'alternative à son usage. L'objectif de ce travail est de déterminer si ces demandes sont compatibles avec les vitesses pratiquées actuellement à l'aide de chacun des modes de transport. Plus précisément notre objectifs est de répondre à ces questions : Est-ce que les conducteurs ont une bonne appréciation de la vitesse quant à leur choix de comportement ? Combien de conducteurs pourraient gagner du temps en changeant de mode de transport ? Que peut-on attendre, en termes de réduction de l'usage de la voiture, d'une amélioration significative de l'offre de transports collectifs ou bien d'une réduction forte de la vitesse de l'automobile ? Cet article présente la méthodologie ainsi que les principaux résultats obtenus à travers les simulations conduites sur les agglomérations de Lyon et de Paris.

La prevision de la mobilite a long terme est un des grands defis de la planification des transports urbains. L’approche demographique, basee sur des enquetes periodiques, rend compte de la dynamique du comportement d’individus appartenant a differentes generations lors des etapes successives de leur cycle de vie. La decomposition des effets temporels en un effet d’âge et un effet de cohorte de naissance permet de tracer un profil-type au cours du cycle de vie : c’est le principe fondamental du modele « âge-cohorte » de projection de la mobilite urbaine a long terme. Le modele est applique a l’agglomeration de Lille, ou nous disposons de trois enquetes espacees de 11 ans.

Bussière (Yves), Armoogum (Jimmy), Madré (Jean-Loup). - Are we Close to Saturation ? A Demographic Approach to the Number of Cars per Household in Three Urban Regions Given the limited range of traditional models which are often based on an assessment of the effects of income at one point in time, we developed a longitudinal approach based on the study of the behaviour of successive generations during their lifetime. The paper provides a synthesis of long-term projections of the motorization of households in the Île-de- France area (11 million inhabitants), the Montreal metropolitan area (3 million) and the Grenoble urban area (0,5 million). The history of motorization extends further back in North America than in Europe, and throws light on the saturation concept (a progressive slowdown towards the asymptotes). The second car becomes the principal source of growth in terms of the number of cars on the road, and urban sprawl leads to the prediction that the number of vehicles on the periphery of an area will increase, whereas growth will be more moderate in the more densely populated areas., Bussière (Yves), Armoogum (Jimmy), Madré (Jean-Loup). - Vers la saturation ? Une approche démographique de l'équipement des ménages en automobile dans trois régions urbaines Partant des limites de la portée dynamique des modèles classiques qui reposent souvent sur une estimation transversale des effets du revenu, nous avons développé une approche longitudinale reposant sur le suivi du comportement des générations successives au cours de leur cycle de vie. Cet article fournit une synthèse des projections à long terme de la motorisation des ménages en Île-de-France (11 millions d'habitants), dans la Région Métropolitaine de Montréal (3 millions) et dans la Région Urbaine de Grenoble (0,5 million). L'histoire de la motorisation, plus ancienne en Amérique du Nord qu'en Europe, éclaire le concept de saturation (ralentissement progressif de la croissance vers des asymptotes déterminées par le modèle et non fixées a priori). La seconde voiture devient partout le principal moteur de la croissance du parc automobile et l'étalement urbain laisse prévoir une véritable explosion du nombre des véhicules en périphérie, alors que sa croissance serait beaucoup plus modérée dans les zones denses., Bussière (Yves), Armoogum (Jimmy), Madré (Jean-Loup). - i Hacia la saturación ? Un análi- sis demográfico del equipamiento automovilístico de los hogares en très regiones ur- banas Partiendo de los limites de alcance dinámico de los modelos clásicos, que a menudo se basan en una estimación transversal de los efectos de renta, los autores elaboran un análisis longitudinal basado en el seguimiento del comportamiento de generaciones sucesi- vas a lo largo de su ciclo de vida. El artículo ofrece una síntesis de las proyecciones a largo plazo de la motorización de los hogares en Ile-de-France (11 millones de habitantes), la región metropolitana de Montreal (3 millones) y la región urbana de Grenoble (0,5 millones). La historia de la motorización, más antigua en America del Norte que en Europa, cla- rifica el concepto de saturación (disminución progresiva del crecimiento hacia las asimpto- tas determinadas por el modelo y no fijadas a priori). El segundo coche es en todos los casos la razón principal del crecimiento del parque automóvil y la expansion urbana hace prever una explosion del numero de vehículos en la periferia ; el crecimiento séria mucho más moderado en las zonas densas., Madré Jean-Loup, Armoogum Jimmy, Bussière Yves. Vers la saturation ? Une approche démographique de l'équipement des ménages en automobile dans trois régions urbaines. In: Population, 51ᵉ année, n°4-5, 1996. pp. 955-977.

This workshop discussed various aspect of the mathematical part of survey methodology, as well as archiving and confidentiality issues aimed at improving data quality and its use through time. Participants identified ways to correct or minimize bias by dealing with incomplete sampling frames, using weighing and imputing procedures. We discussed methods to archive and share GPS-based survey data to preserve anonymity. Finally, we debated research needs on these topics for the next following years.

La disparition du retour au domicile pendant la pause méridienne résulte de changements sociétaux profonds et se traduit par des modifications des schémas d'activités dont les conséquences sur les déplacements sont mesurables. Peut-on connaître plus précisément l'effet statistique sur la mobilité de cette évolution des pratiques et aussi de son inégale persistance dans le territoire français ? Cet article vise à apporter quelques premières réponses à cette question dont l'enjeu est une meilleure interprétation des données au service des politiques de mobilité.

Les données sur la mobilité jouent un rôle crucial dans la recherche sur le comportement des voyageurs et la prévision de la demande. Le recours exclusif aux techniques de collecte de données conventionnelles, à savoir entretien individuel, entretien téléphonique / Web / personnel assisté par ordinateur, sondage postal et courrier électronique présente de nombreux inconvénients, notamment une lourde charge pour les répondants, d'où un taux de réponse faible, description de la mobilité sur une seule journée par personne, manque de fiabilité en raison des limites de la mémoire humaine, coût élevé vue la nécessité d'un travail intensif, écart important entre les enquêtes périodiques sur la mobilité des ménages, sans oublier les difficultés à combiner et à harmoniser les données des enquêtes dans différents pays ou régions. L'utilisation maintenant sans contrainte du système de positionnement global (GPS) a ouvert de grandes possibilités pour améliorer la qualité de ces données. Les journaux relevés GPS sont objectifs, nombreux abondants, continus, détaillés et précis d'un point de vue spatio-temporel. Cependant, les informations de positionnement en elles-mêmes ne sont pas éligibles suffisantes pour l'analyse de la mobilité en raison de l'absence de caractéristiques des déplacements. Cette lacune a entraîné un développement substantiel de deux nouveaux domaines de recherche: l'imputation des modes de transport déplacement et du motif du voyage à partir des données GPS, respectivement. À l'heure actuelle, les téléphones intelligents sont les appareils privilégiés pour collecter les traces GPS et la description des déplacements correspondants. La détection du mode et du motif est une étape essentielle avant de procéder à une analyse du comportement de déplacement (choix du mode ou temps passé dans les activités, par exemple). En ce sens, les performances des algorithmes d'inférence du mode et du motif déterminent le potentiel de l'utilisation d'enquêtes basées sur GPS en tant que complément, voire même alternative complète aux techniques conventionnelles. Dans la littérature, il existe trois lacunes dans la recherche concernant les algorithmes d'imputation et pour les enquêtes assistées par GPS. La première provient du fait que, jusqu'à présent, les enquêtes GPS ont été menées essentiellement dans les zones urbaines des pays développés et parfois aussi en Chine. Par conséquent, l'utilisation du GPS dans les enquêtes sur la mobilité dans les villes des pays en développement reste à développer. La seconde et la troisième lacunes sont les conséquences de cette première limitation. La liste des modes de détection comprend la marche, le vélo, le transport en commun et la voiture, mais pas la moto, l'un des principaux moyens utilisés dans les pays émergents. Enfin, les imputations ont été mises en oeuvre très fréquemment avec le support des Systèmes d'Information Géographique (SIG); Cependant, les données SIG ne sont pas partout disponibles et d'une qualité suffisante. L'absence de solutions raisonnables pour imputer les caractéristiques des déplacements à partir des données GPS sans l'apport d'un SIG constitue une lacune de la recherche. Cette thèse a pour objectif de rechercher des réponses aux trois questions susmentionnées en construisant à la fois des modèles d'inférence de mode et de motif. Deux ensembles de données ont été utilisés. Le premier a été collecté en Rhône-Alpes, en France, par un appareil dédié, tandis que le second a été rassemblé à Hanoï, au Vietnam, par smartphone. Sur la base des performances des classificateurs des recommandations visant à améliorer la qualité des enquêtes par GPS en général et pour les pays en développement en particulier ont été proposées. Mobility data play a crucial role in travel behavior research and demand forecast. The complete reliance on conventional datum collection techniques, that is, face-to-face interview, computer-assisted telephone/web/personal interview, postal survey and email has a number of big drawbacks, including (1) high burden on respondents, thus high non-response rate, (2) inclusion of one-day data per person, (3) lack of reliability due to human memory limits and habit of rounding travel time, (4) high cost with intensive labor and (5) big time gap between periodic household travel surveys not to mention the difficulties in combining and harmonizing data of surveys in different regions or countries. The unlimited use of Global Positioning System (GPS) has opened up great opportunities for dealing with the problem of poor data. GPS logs are objective, numerous, continuous, detailed and accurate spatiotemporally. Yet, positioning information itself is not eligible for analysis due to the lack of trip characteristics. This deficiency has induced the substantial development of two new research fields that are involved in imputing transportation modes and trip purposes from GPS data, respectively. On-board devices were initially utilized to take advantage of electricity. Afterwards, lightweight, small and wearable personal devices have been developed to collect data at person level, which emphasized the need of detecting trip modes. Currently, smartphone is the most preferred devices to gather both logs and their corresponding so-called ground truth. Detections of mode and purpose are essential steps prior to do any travel behavior analyses (e.g. mode choice or time spending in activities). In this sense, the performances of mode and purpose inference algorithms determine the potential of employing GPS-based surveys as a supplement and even an entire alternative to conventional techniques. In the literature, there are three research gaps related to imputation algorithms and GPS-assisted surveys. The first is the great focus of investigations in well-structured urban areas of developed countries and occasionally in China. Therefore, the use of GPS in mobility surveys in cities of developing countries has been questionable. The others are consequences of the first limitation. Second, the list of mode detection encompasses walk, bike, transit and car but not motorcycle that is one of the main means in emerging countries. Last, purpose imputation has been implemented very frequently with the support of GIS data; however, GIS data are not available and good enough everywhere. Lack of reasonable solutions to derive purposes from GPS data without GIS data is a gap. This thesis aims at seeking answers to three mentioned-above existing questions by building both mode and purpose inference models. Two data sets were used. The first was collected in Rhone-Alpes, France by dedicated device whilst the second was gathered in Hanoi, Vietnam by smartphone. Based on prediction results, the discussions and recommendations for enhancing the quality of GPS-based surveys in general and for developing countries in particular have been proposed.

Travel survey methods based on new technologies have evolved in the past few decades, shifting from limited experiments to large-scale travel surveys. GPS-based data collection methods have become particularly popular in travel behavior research, mainly because of the worldwide coverage and the accuracy of the GPS system. We have taken the opportunity of the French National Travel Survey (FNTS) to have the first nationwide experience with embedding such a 'GPS package' in a traditional survey, with a sub-sample of approximately 957 voluntary respondents. This thesis begins by presenting a review of interest in new technologies for mobility surveys, advantages and limitations. Prior to going into the field with the processing of GPS data, it is important to understand the likely acceptability of GPS device among FNTS respondents - specifically whether a sufficient proportion would be willing to use a GPS device while completing a travel diary and, if not, why not. The findings confirmed that the voluntary respondents with GPS receiver are more conscientious and better describe their mobility. We find the profile of respondents who accept the GPS survey. One of the main reason for low GPS acceptance rate in the FNTS 2007 is the availability of GPS. For increasing the GPS availability rate, we need to calculate the time and duration of the survey visits to reduce the GPS immobilization rate. A challenge in the GPS data post-processing is the development of methods to fill GPS missing data and to reconstitute automatically continuous sequence, both in space and time. Then, the architecture of GPS data post-processing for the FNTS 2007-08 is described. After, the results of the post processing software is calibrated by comparison with conventional methods, such as CAPI-GPS providing for a few trips additional information on the reliability of the device and on more detailed characteristics (e.g. mode and purpose). Next, we provide an overview of comparing the descriptions of mobility obtained by two methods CAPI and GPS in the days covered by these two observation tools. The automatic matching of GPS traces to daily trips based on their chronologies and the manual control of this matching are performed. By identifying the characteristics of unmatched GPS traces and daily trips, we expect the reasons for their mismatches. Finally, we evaluate the contributions and challenges of data collection using GPS device. This study shows that the GPS survey can be used successfully to complete the conventional transport surveys, but it is still too early to predict the complete substitution of conventional survey by the GPS mobility survey.; Les méthodes de collecte de données sur la mobilité basée sur les nouvelles technologies ont évolué au cours des dernières décennies. Le suivi de la mobilité par GPS pour la recherche sur les comportements de déplacement se diffuse, principalement en raison de la couverture mondiale et de la précision du système GPS. Le sous-échantillon d'environ 957 volontaires qui ont accepté de porter un GPS pendant au moins une semaine dans le cadre de l'Enquête Nationale Transport et Déplacements (ENTD) 2007-08 est la première expérience nationale de ce type dans le monde. Cette thèse présente l'intérêt des nouvelles technologies pour les enquêtes de mobilité, leurs avantages et leurs biais. D'abord, en étudiant l'acceptabilité et les biais des enquêtes de mobilité à l'aide de GPS, les résultats confirment que les enquêtés volontaires pour une enquête GPS sont plus consciencieux et décrivent mieux leur mobilité. Ensuite, nous décrivons le profil des enquêtés qui acceptent l'enquête GPS. L'une des raisons principales du faible taux de l'acceptation du GPS dans l'ENTD 2007-2008 est la disponibilité de GPS. Donc pour l'accroissement du taux de disponibilité des GPS, nous avons besoin de calculer les temps et durées des visites de l'enquête pour réduire le taux d'immobilisation forcé des GPS. Indépendamment de ce problème d'acceptabilité du GPS par l'enquêté, un défi dans le traitement a posteriori des données collectées par GPS est la mise au point de méthodes permettant de combler les données manquantes et de reconstituer de manière automatisée des séquences continues, à la fois dans l'espace et dans le temps. Nous présentons l'algorithme de post-traitement et les résultats du logiciel de post-traitement des données GPS de l'ENTD 2007-2008. La validation est effectuée en comparant avec les données collectées par les méthodes classiques, ainsi qu'avec le CAPI-GPS qui fournit des informations supplémentaires sur la fiabilité de l'appareil et des caractéristiques plus détaillées (mode, motif...) pour les déplacements réalisés un jour tiré au sort. Ensuite, nous comparons les descriptions de la mobilité obtenues par deux méthodes, questionnaire classique de l'enquête, d'une part et traces GPS d'autre part, pour les jours couverts par ces deux instruments d'observation. Un panorama de l'appariement des traces GPS et des déplacements quotidiens est effectué en s'appuyant sur leur chronologie de manière automatique mais également par un contrôle manuel. En identifiant les caractéristiques des déplacements quotidiens et traces GPS non appariés, nous estimons les raisons de leur non-appariement. Cette recherche montre que l'enquête à l'aide de GPS peut être utilisée avec succès pour compléter les enquêtes de transport classiques, mais qu'il est encore prématuré d'imaginer la substitution complète des enquêtes classiques de mobilité par l'enquête GPS.

Travel survey methods based on new technologies have evolved in the past few decades, shifting from limited experiments to large-scale travel surveys. GPS-based data collection methods have become particularly popular in travel behavior research, mainly because of the worldwide coverage and the accuracy of the GPS system. We have taken the opportunity of the French National Travel Survey (FNTS) to have the first nationwide experience with embedding such a 'GPS package' in a traditional survey, with a sub-sample of approximately 957 voluntary respondents. This thesis begins by presenting a review of interest in new technologies for mobility surveys, advantages and limitations. Prior to going into the field with the processing of GPS data, it is important to understand the likely acceptability of GPS device among FNTS respondents - specifically whether a sufficient proportion would be willing to use a GPS device while completing a travel diary and, if not, why not. The findings confirmed that the voluntary respondents with GPS receiver are more conscientious and better describe their mobility. We find the profile of respondents who accept the GPS survey. One of the main reason for low GPS acceptance rate in the FNTS 2007 is the availability of GPS. For increasing the GPS availability rate, we need to calculate the time and duration of the survey visits to reduce the GPS immobilization rate. A challenge in the GPS data post-processing is the development of methods to fill GPS missing data and to reconstitute automatically continuous sequence, both in space and time. Then, the architecture of GPS data post-processing for the FNTS 2007-08 is described. After, the results of the post processing software is calibrated by comparison with conventional methods, such as CAPI-GPS providing for a few trips additional information on the reliability of the device and on more detailed characteristics (e.g. mode and purpose). Next, we provide an overview of comparing the descriptions of mobility obtained by two methods CAPI and GPS in the days covered by these two observation tools. The automatic matching of GPS traces to daily trips based on their chronologies and the manual control of this matching are performed. By identifying the characteristics of unmatched GPS traces and daily trips, we expect the reasons for their mismatches. Finally, we evaluate the contributions and challenges of data collection using GPS device. This study shows that the GPS survey can be used successfully to complete the conventional transport surveys, but it is still too early to predict the complete substitution of conventional survey by the GPS mobility survey., Les méthodes de collecte de données sur la mobilité basée sur les nouvelles technologies ont évolué au cours des dernières décennies. Le suivi de la mobilité par GPS pour la recherche sur les comportements de déplacement se diffuse, principalement en raison de la couverture mondiale et de la précision du système GPS. Le sous-échantillon d'environ 957 volontaires qui ont accepté de porter un GPS pendant au moins une semaine dans le cadre de l'Enquête Nationale Transport et Déplacements (ENTD) 2007-08 est la première expérience nationale de ce type dans le monde. Cette thèse présente l'intérêt des nouvelles technologies pour les enquêtes de mobilité, leurs avantages et leurs biais. D'abord, en étudiant l'acceptabilité et les biais des enquêtes de mobilité à l'aide de GPS, les résultats confirment que les enquêtés volontaires pour une enquête GPS sont plus consciencieux et décrivent mieux leur mobilité. Ensuite, nous décrivons le profil des enquêtés qui acceptent l'enquête GPS. L'une des raisons principales du faible taux de l'acceptation du GPS dans l'ENTD 2007-2008 est la disponibilité de GPS. Donc pour l'accroissement du taux de disponibilité des GPS, nous avons besoin de calculer les temps et durées des visites de l'enquête pour réduire le taux d'immobilisation forcé des GPS. Indépendamment de ce problème d'acceptabilité du GPS par l'enquêté, un défi dans le traitement a posteriori des données collectées par GPS est la mise au point de méthodes permettant de combler les données manquantes et de reconstituer de manière automatisée des séquences continues, à la fois dans l'espace et dans le temps. Nous présentons l'algorithme de post-traitement et les résultats du logiciel de post-traitement des données GPS de l'ENTD 2007-2008. La validation est effectuée en comparant avec les données collectées par les méthodes classiques, ainsi qu'avec le CAPI-GPS qui fournit des informations supplémentaires sur la fiabilité de l'appareil et des caractéristiques plus détaillées (mode, motif...) pour les déplacements réalisés un jour tiré au sort. Ensuite, nous comparons les descriptions de la mobilité obtenues par deux méthodes, questionnaire classique de l'enquête, d'une part et traces GPS d'autre part, pour les jours couverts par ces deux instruments d'observation. Un panorama de l'appariement des traces GPS et des déplacements quotidiens est effectué en s'appuyant sur leur chronologie de manière automatique mais également par un contrôle manuel. En identifiant les caractéristiques des déplacements quotidiens et traces GPS non appariés, nous estimons les raisons de leur non-appariement. Cette recherche montre que l'enquête à l'aide de GPS peut être utilisée avec succès pour compléter les enquêtes de transport classiques, mais qu'il est encore prématuré d'imaginer la substitution complète des enquêtes classiques de mobilité par l'enquête GPS.