Article

Dynamic Choice Set Generation Based on Global Positioning System Trajectories and Stated Preference Data

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

A method that generated choice sets for commuters in the Munich, Germany, metropolitan area was explored. The method used Global Positioning System trajectories and interview data from 300 commuters over an 8-week survey to combine chosen, known, and generated routes into choice sets for route choice modeling. The method used revealed preference routes as well as stated preference routes to calculate accepted detour factors, which were then used as boundary conditions for choice set generation using path enumeration. On the basis of a spatial choice set, the method generated time-dependent choice sets by attributing all routes with actual travel times at the time traveled.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... Clustering is based on geometric properties of time series of GPS recordings. Many researchers (e.g Giannotti et al. 2007;Zheng et al. 2010;Yan and Spaccapietra 2009;Andrienko et al. 2011;Furletti et al. 2013) focus on trajectory annotation [mining of travel behaviour, e.g. in order to establish route choice sets (Pillat et al. 2011)]. Kim and Mahmassani (2015) use DBSCAN to find traffic stream clusters and define the concept of a cluster representative subsequence. ...
Article
Full-text available
We propose a method, whose purpose is to combine a set of GPS traces collected by bicyclists with a set of notifications of problematic situations to determine an optimal action plan for solving safety related problems in a traffic network. In particular, we use optimization to determine which problem locations to resolve under a given budget constraint in order to maximize the number of impediment free trips. The method aims to suggest a priority of impediments to resolve, which would be manually infeasible. The proposed method consists of two steps. First, problematic locations are clustered, where each cluster corresponds to a so-called impediment. Each impediment is associated with trips nearby using a distance function. The trip set is partitioned by matching each trip with the largest set of its affecting impediments. Solving all impediments associated with such a part induces a cost and makes the associated part of trips impediment free. The second step aims to find the set of impediments that can be solved with a given budget and that makes the maximum number of trips impediment free. A branch-and-bound optimizer for the second step is presented and evaluated. The clustering parameters affect the set of identified impediments and the extent of each of them. In order to evaluate the sensitivity of the result to the clustering parameters a technique is proposed to consistently estimate the impediment resolution cost. Our study aims to support the interactive urban designer to improve the urban bicycle road infrastructure. By providing a method to prioritize between impediments to resolve, it also aims to contribute to a safer and more attractive traffic situation for bicyclists.
... Clustering is based on geometric properties of time series of GPS recordings. Many researchers [7,21,20,1,6] focus on trajectory annotation (mining of travel behaviour e.g. in order to establish route choice sets [15]). The authors of [9] use DBSCAN to find traffic stream clusters and define the concept of cluster representative subsequence. ...
Article
Full-text available
A set of GPS traces for bicyclists and a set of notifications by bicyclists of problematic situations (spots identified by GPS records) had been collected independently. The data collection periods did not coincide but overlapped and none was contained in the other one. The aim is to use both datasets to determine an optimal action plan for problem solving given a limited budget. First, problematic locations are clustered. Each cluster corresponds to an impediment. Impediments are then associated with trips using a distance function. The aim is to find out which impediments to solve under a given budget constraint in order to maximize the number of impediment free trips. Thereto the trip set is partitioned by matching each trip with the largest set of its affecting impediments. Solving all impediments in such set induces a cost and makes the associated part of trips impediment free. An optimizer is presented and evaluated.
... Empirical research on route choice behaviour shed light on the attributes influencing the route selection of drivers (Abdel-Aty et al., 1997;Arnott et al., 1992;Gao et al., 2010;Wu et al., 2018;Xu et al., 2011). Discrete choice modelling is the the most frequently used methodology in the literature to unravel the route choice behaviour of individuals (Dalumpines and Scott, 2017;Emmerink et al., 1996;Habib et al., 2013;Pillat et al., 2011;Prato, 2009;Tawfik and Rakha, 2013). The identified factors can be categorized into route attributes, traffic information, trip specifications, drivers' attitudes and socio-demographic characteristics. ...
Article
Full-text available
Electric travelling appears to dominate the transport sector in the near future due to the needed transition from internal combustion vehicles (ICV) towards Electric Vehicles (EV) to tackle urban pollution. Given this trend, investigation of the EV drivers' travel behaviour is of great importance to stakeholders including planners and policymakers, for example in order to locate charging stations. This research explores the Battery Electric Vehicle (BEV) drivers route choice and charging preferences through a Stated Preference (SP) survey. Collecting data from 505 EV drivers in the Netherlands, we report the results of estimating a Mixed Logit (ML) model for those choices. Respondents were requested to choose a route among six alternatives: freeways, arterial ways, and local streets with and without fast charging. Our findings suggest that the classic route attributes (travel time and travel cost), vehicle-related variables (state-of-charge at the origin and destination) and charging characteristics (availability of a slow charging point at the destination, fast charging duration, waiting time in the queue of a fast-charging station) can influence the BEV drivers route choice and charging behaviour significantly. When the state-of-charge (SOC) at the origin is high and a slow charger at the destination is available, routes without fast charging are likely to be preferred. Moreover, local streets (associated with slow speeds and less energy consumption) could be preferred if the SOC at the destination is expected to be low while arterial ways might be selected when a driver must recharge his/her car during the trip via fast charging.
... Discrete choice modeling, which is generally based on expected utility theory, has been widely applied to analyze route choice behaviors (Broach et al. 2012;Hood et al. 2013;Pillat et al. 2011). The route choices of EV drivers depend on some of the same main factors as those of GV drivers, including individual characteristics, route attributes, and traffic information. ...
Article
The adoption of battery electric vehicles (BEVs) is widely expected to help improve air quality in Chinese cities, whereas this adoption may change traffic status related to vehicle energy consumption and emissions. This paper presents a traffic status-sensitive approach to evaluating the environmental effects and accounting for the potential impacts of BEVs on traffic flows. First, a multinomial logit (MNL) BEV route choice model is estimated, and the different route choice preferences between BEV and gasoline vehicle (GV) drivers are analyzed. Second, a stochastic assignment approach for mixed electric and gasoline vehicular flows is developed, incorporating BEV and gasoline vehicle (GV) route choice models. Finally, using actual vehicle trip data, the proposed approach is applied to calculate the traffic status and evaluate the environmental effects under various BEV penetrations in Beijing. The results indicate that the relative decreases in emissions are less than the BEV penetration rate, and increasing BEVs appears to change network traffic status and then increase the emission rates of GVs, especially during peak hours.
... However, due to the assumption that the error component follows the identical and independent Gumbel distribution (IID), the biases exist in parameter estimations in most cases when facing the overlapping problem between alternatives in transportation networks [13]. As a consequence, a host of revised models were proposed to overcome the defect brought by the independence of irrelevant alternatives (IIA) property, mainly referring to overlapping problem in route choice, such as C-logit, path sized logit (PSL), and cross-nested logit (CNL), etc. [14][15][16][17][18][19]. It is worth noting that the overlapping issue also exists in the Metro network, which is easy to be neglected in route choice modeling [20]. ...
Article
Full-text available
With the rapid development of the Metro network in China, the greatly increased route alternatives make passengers’ route choice behavior and passenger flow assignment more complicated, which presents challenges to the operation management. In this paper, a path sized logit model is adopted to analyze passengers’ route choice preferences considering such parameters as in-vehicle time, number of transfers, and transfer time. Moreover, the “perceived transfer threshold” is defined and included in the utility function to reflect the penalty difference caused by transfer time on passengers’ perceived utility under various numbers of transfers. Next, based on the revealed preference data collected in the Guangzhou Metro, the proposed model is calibrated. The appropriate perceived transfer threshold value and the route choice preferences are analyzed. Finally, the model is applied to a personalized route planning case to demonstrate the engineering practicability of route choice behavior analysis. The results show that the introduction of the perceived transfer threshold is helpful to improve the model’s explanatory abilities. In addition, personalized route planning based on route choice preferences can meet passengers’ diversified travel demands.
... A similar technique is proposed by Pillat et al. (2011) who use a detour threshold that is a function of the duration of the first part of the route being constructed. The function was derived from routes collected by a survey. ...
... Multinomial logit (MNL) model, in which very strict assumption of independence of irrelevant alternatives (IIA) is needed, has been widely applied to the analysis of route choice behavior during past decades. Several modifications or generalizations of the logit structure (e.g., C-logit, path-size logit (PSL), cross-nested logit (CNL)) have been proposed to relax the IID assumptions of the error term of MNL model, and these models have been widely used in many studies (Pillat et al., 2011;Hood et al., 2011;Broach et al., 2012;Train, 2001;Koppelman and Wen, 2000). ...
Preprint
Full-text available
Several route choice models developed in the literature were based on a relatively small number of observations. With the extensive use of tracking devices in recent surveys, there is a possibility to obtain insights with respect to the traveler's choice behavior. In this paper, different path generation algorithms are evaluated using a large GPS trajectory dataset. The dataset contains 6,000 observations from Tel-Aviv metropolitan area. An initial analysis is performed by generating a single route based on the shortest path. Almost 60% percent of the 6,000 observations can be covered (assuming a threshold of 80% overlap) using a single path. This result significantly contrasts previous literature findings. Link penalty, link elimination, simulation and via-node methods are applied to generate route sets, and the consistency of the algorithms are compared. A modified link penalty method, which accounts for preference of using higher hierarchical roads, provides a route set with 97% coverage (80% overlap threshold). The via-node method produces route set with satisfying coverage, and generates routes that are more heterogeneous (in terms number of links and routes ratio).
Thesis
Full-text available
A framework of travel demand modeling is devised in this work that uses cell-phone network generated data to estimate the operational demand matrices in fine spatial and temporal resolution. Due to its high penetration rate, extensive coverage and ubiquitous use, opportunistically collected cell-phone data has a great potential to be used for passive probing of traffic. For megacities like Dhaka, this method has definite advantage over others because of lack of permanent data collection infrastructure, high market penetration rate (over 90%) of cell-phones, spatially densified network structure and less expense involved in the process. Also, from a questionnaire survey done for this thesis it is found that 80% of mobile phone users make six or more calls per day and 28% of them make more than 10 calls during their trip. Motivated by the prospect, a large dataset containing one month call data records (CDR) of nearly 1.3 billion network connections for about 5 million subscribers was collected from the largest network operator of Bangladesh (Grameen Phone). To the knowledge of the author, this is the largest billing dataset used in this purpose. Computationally efficient and simplified algorithms are designed to process and analyze this big data of 80 GB. A part of the simplification is achieved by introducing temporal resolution and filtering out infrequent callers. This simplification has reduced the computation time by 35 times. The database is processed to impute trips and synthesize time-dependent trip matrices. Derived matrices are important indicator of travel pattern or characteristics of persontrips. In order to convert these ‘pattern’ or ‘seed’ matrices to operational vehicle-trip matrices, scaling factors for different time periods are developed. Surveyed commute flow data (DHUTS, 2010) is used to derive these factors. The factors are determined separately for each OD pair due to variations in person-trip to vehicle-trip conversion factors, operator subscription, network coverage, market penetration rate and type of technology used. Also, clustering of matrices for identical time periods is done to get representative trip matrices for those periods. Therefore, devised framework is capable of incorporating more dataset and producing more reliable factors for similar time period. ‘Seed’ matrices for that corresponding time periods are factored and the factored matrices are assigned in modeling software, TransCAD, using prevalent demand modeling techniques. A validation framework is presented at the end to find out the accuracy of our methodology. Video recording made during the time period was used as ‘ground truth’ data. But due to unavailability of this data for the entire modeled period and network, some complementary data from secondary source was used. The result shows less than 40 percent deviation in network flows for 85 percent links of the network that is relatively less compared to other demand modeling practices where it is common to get greater percentage of deviation. Additional information regarding users’ behavior in the network is also extracted. The dynamic calling pattern map shows us population distribution and activity-space over time. It is found that the number of calls a user make in a day follows a diminishing distribution pattern. The logarithm of inter-call time follows a quadratic frequency distribution. Cumulative distribution function for the number of imputed trips by a user can be well approximated by a normal distribution or an exponential distribution. The relationship between tripmaking and callmaking of each user reveals that a minimum number of callmaking is required to sense certain number of trips for a specific user. Also, most of the data points (99.5%) are contained within the limit of 20 trips per day. Number of trip sensed is at the maximum for Thursdays and at the minimum for Fridays. Moreover, an absence of morning peak is found for weekends and peak-spreading of morning peak is observed for weekdays.
Article
Analysis of GPS traces shows that people often do not use the least cost path through the transportation network while making trips. This leads to the question which structural path characteristics can be used to construct realistic route choice sets for use in traffic simulation models. In this paper, we investigate the hypothesis that, for utilitarian trips, the route between origin and destination consists of a small number of concatenated least cost paths. The hypothesis is verified by analyzing routes extracted from large sets of recorded GPS traces which constitute revealed preference information. Trips have been extracted from the traces and for each trip the path in the transportation network is determined by map matching. This is followed by a path decomposition phase for which the algorithm constitutes the first contribution of this paper. There are multiple ways to split a given path in a directed graph into a minimal number of subpaths of minimal cost. By calculating two specific path splittings, it is possible to identify subsets of the vertices (splitVertexSuites) that can be used to generate every possible minimum path splitting by taking one vertex from each such subset. As a second contribution, we show how the extracted information is used in microscopic travel simulation. The distribution for the size of the minimum decomposition, extracted from the GPS traces, can be used in constrained enumeration methods for route choice set generation. The sets of vertices that can act as boundary vertices separating consecutive route parts contain way points (landmarks) having a particular meaning to their user. The paper explains the theoretical aspects of route splitting as well as the process to extract splitVertexSuites from big data. It reports statistical distributions extracted from sets of GPS traces for both multimodal person movements and unimodal car trips.
Thesis
Full-text available
Intelligent traffic management has been the means to utilize existing capacities on overloaded road networks for quite some time. In recent years traffic information has become more and more important. The increasing popularity of navigation systems as well as the growing availability of current travel times makes it possible to provide a vast number of car drivers with traffic state information via onboard navigation devices or dynamic roadside traffic signs. The first part of this work analyses drivers’ acceptance of traffic information and compliance with route guidance today on the basis of an empirical study of almost 300 car commuters in the greater Munich area over a period of eight weeks. An analysis of the collected data reveals that traffic information is not only used for route guidance in case of needing directions but it is also used rather frequently on everyday trips, such as from home to work. Today, the level of drivers’ information on potential alternative routes and current travel times is based on incomplete as well as temporally delayed information and thus is greatly dependent on the driver’s own knowledge of historical daytime-dependent travel times in the road network. In the second part of this work econometric choice models are presented based on the collected GPS trajectories as well as questionnaire data using Maximum-Likelihood estimation. These choice models identify the variables influencing route and departure time choice. Special focus is given to the influence of traffic reports received via radio, route guidance via navigation devices or via dynamic roadside traffic signs, traffic state information displayed as Level-of-Service map on navigation devices, and travel time information on dynamic roadside traffic signs. The analysis of over 16,000 trips shows that survey participants have a strong preference towards their usual main route. Besides the current travel time, traffic information via radio or via Level-of-Service map has the strongest influence on the choice to divert from the main route. The probability of diverting to an alternative route is equal for a 5-minute increase in travel time as for 6 kilometres of congestion or 10 kilometres of minor delays when the information is broadcast over the radio or displayed through the Level-of-Service map. The evaluation of the questionnaire shows the influence of current travel time information on departure time choice. The results greatly depend on the flexibility of the working hours of the decision maker. For a usual commuting distance of 30 kilometres, persons with flexible working hours show a willingness to change their usual departure time by 15 minutes for a 10-minute travel time reduction, whereas persons with fixed working hours are willing to change their departure time for a 25-minute travel time reduction. The third part of this work includes modelling concepts to determine the potential of traffic information to reduce the transport time expenditure as well as fuel consumption in the entire survey area. Therefore, a comparison of today’s traffic situation with a state with perfect information and a system-optimal state from the transport planning perspective is performed. The separate models of route and departure time choice are integrated within a macroscopic traffic assignment model. Optimization of route choice by providing all drivers with perfect information on current travel times reduces the transport time expenditure by 4% in the morning peak hour. A substantially larger reduction of almost 8%, and thus close to a system-optimal state, is achieved by perfect information on the optimal departure time which causes a temporal distribution of the travel demand. The empirical analysis within this work and the route and departure time choice models based thereon determine impacts of traffic information in the study area of Munich today as well as possibly in the future. The results show the potential of traffic information for the optimization of traffic flows in road networks. Furthermore, this work provides an important contribution to model-based decision guidance for practitioners in optimizing traffic management strategies as well as for politicians deciding on future investments in information infrastructure and technologies.
Article
Transportation systems are inherently uncertain because of random disruptions; nevertheless, real-time Information can help travelers make better route choices under such disruptions. The first revealed-preference study of routing policy choice Is presented. A "routing policy" is defined as a decision rule applied at each link that maps possible realized traffic conditions to decisions to be made on the link next. The policy represents a traveler's ability to incorporate real-time information not yet available at the time of decision. Two case studies are conducted in Stockholm, Sweden, and in Singapore. Data for the underlying stochastic time-dependent network are generated from taxi GPS traces through map-matching and nonparametric link travel time estimation. An efficient algorithm to find the optimal routing policy in large-scale networks is first presented, which is a building block of any routing policy choice set generation method. The routing policy choice sets are then generated by link elimination and simulation. The generated choice sets are first evaluated on the basis of whether they include the observed traces on a specific day, or coverage. The sets are then evaluated on the basis of "adaptiveness," defined as the capability of a routing policy to be realized as different paths over different days. A combination of link elimination and simulation methods yields satisfactory coverage. The comparison with a path choice set benchmark also suggests that a routing policy choice set could potentially provide better coverage and capture the adaptive nature of route choice.
Article
The extensive use of GPS-based travel surveys in the past few years now allows vehicle movements to be traced and, thus, data on the actual routes chosen for various trips to be collected. However, efforts on the empirical modeling of route choices through the use of GPS traces are still limited. In this context, the broad focus of this research was to combine data from a large-scale GPS-based travel survey and geographic information system-based roadway network databases to develop models for route choice. Data from GPS streams for 1,913 trips were used in this analysis. Three models that considered choice set sizes of five, 10, and 15 alternatives were built. The estimation results indicated statistically significant and intuitively reasonable effects of free-flow travel time, left turns, right turns, intersections, and circuity on the attractiveness of different route alternatives. Furthermore, the sensitivity to these factors was found to vary on the basis of trip (purpose, time of day, and day of the week) and traveler (gender, age, and length of stay at the current home) characteristics.
Article
To track a vehicle using GPS data, map matching is not only to select the map feature that the vehicle was at a GPS point, but also to find out the route it traveled. For GPS data at high polling rate, it is very obvious that the vehicle stays on the same map feature or a map feature directly connected with the previous GPS point. For low polling rates, a map feature where a vehicle is at a GPS point, is not always directly connected with the map feature where the vehicle is at the previous GPS point. In this paper, an improved model is proposed. The candidate set generation considers the difference of the width of links, and the matched map feature selection considers the length of the route, the difference between the azimuth of the GPS point and the map feature and the deviation of the raw GPS position to the candidate map feature. A field test shows that the improved model performs better than the previous model.
Article
Full-text available
An algorithm to solve explicitly the path enumeration problem is proposed. This algorithm is based on the branch-and-bound technique and belongs to the class of deterministic methods along with existing approaches that combine heuristic or randomization procedures with shortest-path search. The branch-and-bound algorithm is formulated, and a methodology is designed for the application of deterministic approaches to a real case study. Path sets generated with different methods are compared for behavioral consistency, namely, the ability to reproduce actual routes chosen by individuals driving habitually from home to work. Choice set compositions for modeling purposes are determined for the consistency of the path generation process with the observed behavior. Further, model estimates and performance for different route choice specifications are examined for both path set compositions. Results suggest that the proposed branch-and-bound algorithm generates realistic and heterogeneous routes, reproduces better the observed behavior of the interviewed drivers, and produces a good choice set for route choice model estimation and performance comparison.
Conference Paper
Full-text available
The paper reports on a method which generates time-space trajectories of travelers by analyzing data from the mobile phone network. The method utilizes mainly the data from location-area-updates, which are recorded from any mobile phone in the standby-mode and thus does not require phone calls or handset equipment. The main idea of the method is to derive location-area-sequences for single trips from the mobile phone data and to compare these with location-area-sequences, which are based on routes through the road and rail network. The method generates a vast number of trajectories on highways and can be used for several applications in the field of transportation planning.
Thesis
Full-text available
Mobilfunkgeräte sind in den letzten Jahren zum ständigen Begleiter der meisten Menschen in Deutschland und in der ganzen Welt geworden. Das Mobilfunknetz kennt während des normalen Betriebs für jedes angeschaltete Mobilfunkgerät die aktuellem ungefähr 30 bis 40 Funkzellen umfassende Location Area. Damit bieten sich die im Betrieb anfallenden Mobilfunkdaten als Datenquelle für Verkehrsingenieure an, um eine große Anzahl von Ortsveränderungen kontinuierlich zu beobachten. Der erste Teil dieser Arbeit beschreibt ein Verfahren, mit dem aus Mobilfunkdaten Trajektorien von Mobilfunkteilnehmern generiert werden können. Diese Methode basiert auf den Daten von Location Area Updates, die immer dann erfolgen, wenn ein Mobilfunkgerät eine Location Area verlässt und in eine andere Location Area wechselt. Diese finden auch dann statt, wenn keine Telefonate geführt werden. Somit können mit dieser Methode die Ortsver¬änderungen aller einge¬schalteten Mobilfunkgeräte im untersuchten Mobilfunknetz erfasst werden. Die Durchführung des Verfahrens mit Mobilfunkdaten von T-Mobile im Untersuchungs-gebiet im Nord-Westen Baden-Württembergs ergibt, dass Trajektorien ab einer Länge von ungefähr 20 Kilometern erzeugt werden können, wobei die Qualität der Trajektorien mit ihrer Länge zunimmt. Damit eignet sich das Verfahren vor allem für Autobahnen und ausgewählte Bundesstraßen, nicht aber für innerstädtischen Verkehr. Im zweiten Teil der Arbeit werden die Trajektorien aus Mobilfunkdaten dazu verwendet, das Routenwahlverhalten mit Maximum-Likelihood-Schätzungen zu analysieren. Dabei wird untersucht, welche Einflussgrößen Verkehrsteilnehmer bei der Routenwahl berücksichtigen. Mögliche Einflussgrößen sind dabei die Routenempfehlungen auf dynamischen Anzeigen entlang der Straße sowie Verkehrs¬meldungen, die über den Verkehrsfunk bzw. das Navigationsgerät empfangen werden. Die Ergebnisse in der Autobahn-Netzmasche zwischen Stuttgart, Karlsruhe, Walldorf und Heilbronn unter Nutzung von ungefähr 1 Million Trajektorien zeigen deutlich, dass die Verkehrsteilnehmer beide genannten Einflussgrößen bei ihren Routenwahlent-scheidungen berücksichtigen. Ungefähr 30 % der Verkehrsteilnehmer auf der Standardroute wechseln bei einer entsprechenden Routenempfehlung auf die Alternativroute. Eine ähnliche Wirkung haben akkumulierte Verkehrs¬meldungen von sieben Kilometer Stau auf der Standardroute. Für den Bereich der Netzbeeinflussungsanlage im Norden von Stuttgart, die den Verkehr zwischen der Autobahn A81 und Stuttgart auf verschiedene Routen steuert, lässt sich dagegen keine Wirkung der Verkehrsmeldungen nachweisen. Zudem ist der Befolgungsgrad der innerhalb des Untersuchungszeitraums von drei Monaten nur selten aktiven Routenempfehlungen auf dynamischen Anzeigen mit 3 % stadtauswärts bzw. 17 % stadteinwärts geringer als in der BAB-Netzmasche. Somit zeigen die Ergebnisse dieser Arbeit, dass Mobilfunkdaten eine sehr interessante Datenquelle für Verkehrsingenieure sind.
Article
Full-text available
Transit assignment procedures need to reflect the constraints imposed by line routes and timetables. They require specific search algorithms that consider transfers between transit lines with their precise transfer times. Such an assignment procedure is presented for transit networks using a timetable-based search algorithm. In contrast to existing timetable-based search methods employing a shortest-path algorithm, the described procedure constructs connections using branch and bound techniques. This approach significantly reduces computing time, thus facilitating the use of timetable-based assignment for large networks. At the same time, it produces better results in cases where slow but cheap or direct connections compete with fast connections that are more expensive or require transfers.
Article
Full-text available
Since the first Global Positioning System (GPS) studies in the mid-1990s, this method of surveying individual travel behavior has gained attention in transport research. Compared with classic travel survey methods, GPS studies offer researchers benefits of more accurate and reliable informa- tion. At the same time, the participants' burden is reduced substantially if the GPS data collection does not involve time-consuming questions. However, without additional information, such as modes and trip pur- poses, extensive postprocessing is required to derive data that can be used for analysis and model estimation. The corresponding procedures are an ongoing research issue. This paper describes a postprocessing procedure needing no input other than the most basic GPS raw data: three-dimensional positions and the corresponding time stamps. First, the data are thoroughly cleaned and smoothed. Second, trips and activities are determined. Third, the trips are segmented into single-mode stages, and the transport mode for each of the stages is identified. The procedure is applied to GPS records collected in the Swiss cities of Zurich, Winterthur, and Geneva. A total of 4,882 participants carried an on-person GPS receiver for an average of 6.65 days. The results are compared with the Swiss Microcensus 2005 to demonstrate that derived data are ready for further applications, such as discrete choice model estimations.
Article
Full-text available
Most route choice models are related to revealed choice behavior and are estimated adding alternative paths to the observed routes. This paper focuses on the effects of choice set composition in route choice modeling by designing an experimental analysis of actual route choice behavior of individuals driving habitually from home to work in an urban network. The numerical analysis concentrates on both a qualitative perspective, by considering path sets built with different generation techniques, and a quantitative perspective, by accounting for path sets constructed with sample size reduction from each initial choice set. Comparison of prediction accuracy across different choice sets suggests that a recently developed branch and bound algorithm generates heterogeneous routes that allow estimating models with better prediction abilities with respect to the outcomes of the drivers’ actual choices. Further, comparison of route choice models across different choice set compositions indicates that non-nested structures, such as C-Logit and Path Size Logit, yield more robust parameter estimates.
Article
Full-text available
This paper proposes an algorithm to explicitly solve the path enumeration problem. This algorithm is based on branch & bound technique and belongs to the class of deterministic methods, alongside existing approaches that combine heuristic or randomization procedures with shortest path search. This paper formulates the branch & bound algorithm and designs a methodology for the application of deterministic approaches to a real case study. Path sets generated with different methods are compared with respect to their behavioral consistency, namely the ability of reproducing actual routes chosen by individuals driving habitually from home to work. Choice set compositions for modeling purposes are determined with respect to the consistency of the path generation process with the observed behavior. Further, model estimates and performances for different route choice specifications are examined for both path set compositions. Results suggest that the proposed branch & bound algorithm generates realistic and heterogeneous routes, reproduces better the observed behavior of the interviewed drivers and produces a good choice set for route choice model estimation and performance comparison.
Article
Full-text available
This paper discusses choice set generation and route choice model estimation for large-scale urban networks. Evaluating the effectiveness of Advanced Traveler Information Systems (ATIS) requires accurate models of how drivers choose routes based on their aware- ness of the roadway network and their perceptions of travel time. Many of the route choice models presented in the literature pay little attention to empirical estimation and validation procedures. In this paper, a route choice data set collected in Boston is described and the ability of several different route generation algorithms to produce paths similar to those ob- served in the survey is analyzed. The paper also presents estimation results of some route choice models recently developed using the data set collected.
Book
Introduction. 1. Transportation Systems. 2. Transportation Supply Models. 3. Random Utility Theory. 4. Transportation Demand Models. 5. Models for Traffic Assignment to Transportation Networks. 6. Intra-Period (Within-Day) Dynamic Models*. 7. Algorithms for Traffic Assignment to Transportation Networks. 8. Estimation of Travel Demand Flows. 9. Transportation Supply Design Models. 10. Transportation Systems Engineering for Planning and Evaluation. A. Review of Numerical Analysis. References. Index. Main Variables.
Article
This article reviews a number of topics related to the modelling and generation of route choice sets, specifically for applications in large networks. It synthesizes existing knowledge using a conceptual framework, and addresses in what respects route choice differ from other travel choices. It shows that it is advantageous to distinguish the processes of choice set formation and choice per se, but also to explicitly separate the modelling steps of choice set generation and choice modelling. The article discusses the various purposes for which route choice sets may be used and what these mean for practical choice set modelling. A generic conceptual scheme is presented relating the distinct key elements of the known route choice set generation approaches aimed at their classification and characterization. Some indications for their empirical validity are presented derived from applications to various uni‐modal and multi‐modal networks.
Article
With the increasing use of GPS in transport surveys, analysts can not only choose from numerous new ways to model travel behaviour but also face several new challenges. For instance, information about chosen routes is now available with a high level of spatial and temporal accuracy. However, advanced post-processing is necessary to make this information usable for route choice modelling. Out of many related issues, this article focuses on the generation of choice sets for car trips extracted from GPS data. The aim is to generate choice sets for about 36,000 car trips made by 2434 persons living in and around Zurich, Switzerland, on the Swiss Navteq network, a very high-resolution network. This network resolution is essential for an accurate identification of chosen routes. However, it substantially increases the requirements for the choice set generation algorithm in regard to performance as well as choice set composition. This article presents a new route set generation based on shortest path search with link elimination. The proposed procedure combines a Breadth First Search with a topologically equivalent network reduction and ensures a high diversity between the routes, as well as computational feasibility for large-scale problems like the one described above. To demonstrate the usability of the algorithm, its performance and the resulting route sets are compared with those of a stochastic choice set generation algorithm.
Article
We consider a graph with n vertices, all pairs of which are connected by an edge; each edge is of given positive length. The following two basic problems are solved. Problem 1: construct the tree of minimal total length between the n vertices. (A tree is a graph with one and only one path between any two vertices.) Problem 2: find the path of minimal total length between two given vertices.
Article
This paper presents a new paradigm for choice set generation in the context of route choice model estimation. We assume that the choice sets contain all paths connecting each origin–destination pair. Although this is behaviorally questionable, we make this assumption in order to avoid bias in the econometric model. These sets are in general impossible to generate explicitly. Therefore, we propose an importance sampling approach to generate subsets of paths suitable for model estimation. Using only a subset of alternatives requires the path utilities to be corrected according to the sampling protocol in order to obtain unbiased parameter estimates. We derive such a sampling correction for the proposed algorithm.Estimating models based on samples of alternatives is straightforward for some types of models, in particular the multinomial logit (MNL) model. In order to apply MNL for route choice, the utilities should also be corrected to account for the correlation using, for instance, a path size (PS) formulation. We argue that the PS attribute should be computed based on the full choice set. Again, this is not feasible in general, and we propose a new version of the PS attribute derived from the sampling protocol, called Expanded PS.Numerical results based on synthetic data show that models including a sampling correction are remarkably better than the ones that do not. Moreover, the Expanded PS shows good results and outperforms models with the original PS formulation.
Article
An efficient computational implementation of a path deletion $K$ shortest paths algorithm and a new algorithm for the same problem are presented. In a path deletion $K$ shortest paths algorithm a sequence $(G_1, G_2, ldots, G_k)$ of networks is defined, such that $G_1$ is the given network and its $k$-th shortest path is trivially determined from the shortest path in $G_k$. In essence, as soon as the shortest path in $G_k$ is determined it is excluded from $G_k$ in such a way that no new paths are formed and no more paths are deleted. So, for each $G_k$, two procedures are executed: a shortest path algorithm and a path deletion algorithm. In the presented computational implementation, all the information resulting from the determination of the $k$-th shortest path is carried throughout $G_k+1, G_k+2, ldots, G_k$. The new algorithm not only keeps this characteristic but also avoids the last $K-1$ executions of a shortest path algorithm, which results in a surprising and very substantial reduction in the execution time. In fact, for randomly generated networks with $10^4$ nodes and $10^5$ arcs, once the shortest path is determined, the new algorithm computes the next $100$ shortest paths in times of the order of $10^-1$ seconds. To illustrate the efficiency of this algorithm, comparative computational experiments are reported.
Article
In this thesis a simultaneous Trip Generation-, Distribution-, Modal-Split and Route Choice Model (modell EVA-U) is elaborated. The model tends to reach a stochastic user equilibrium. The route choice algorithms are not longer part of an assignment procedure but part of the demand model. A consistent assessment of properties of all transport systems is possible. The simultaneous model EVA-U is an advancement of the EVA-Model by Lohse. The model EVA-U is to be assigned to the generalised logit-models. All matrix constrains are taken into account. The assessment is effected by generalised costs. The dependence of routes is taken into account. Moreover, the integration of day time and the schedules of private transport lines is possible. Furthermore, it is possible to integrate a model of parked cars and circuits of inter-modal traffic forms (park and ride) in the Model EVA-U.
Article
Models of urban traveler route choice are reviewed in the context of Intelligent Transportation Systems, particularly Advanced Traveler Information S ystems. Existing models suffer from assumptions of perfect information about travel conditions a nd infinite information processing capabilities of drivers. We present evidence that a majority of travelers fail to minimize travel time or distance. We also show that travelers with more network knowledge appear to vary their commute route to respond to changing travel conditions. Coefficient estimates of a model of network knowledge, based on the geographical idea of spatial ability, are presented. To better understand habitual route choice behavior, we examine many possible route generation algorithms. A simulation approach is preferred because it allows for heterogeneity in driver perceptions and it has a quick computational time. Alternative route choice model specifications such as Multinomial Logit, C-Logit, Path Size Logit, Cross-Nested Logit and Logit Kernel Probit are evaluated. The exponential specification of the Path S ize term, using a large parameter value, offers a considerable improvement in fit over MNL, C -Logit and CNL. A hybrid Path Size Logit and Logit Kernel Probit model offers the best overall fit; however, the stability of these estimates requires further examination. The hybrid Path S ize Logit and CNL model provides the next best empirical fit. Random coefficient specifications of MNL, PS L and LK Probit models were also examined.
Article
This paper deals with algorithms for finding the constrained K-shortest paths (CKSP) and their application to the path enumeration problem. An attractive property of using Constrained Shortest Paths for path enumeration is that paths can be selected based on objective criteria. The conventional way of finding these paths is to compute a sufficiently large number of overall shortest paths, and deleting the ones that do not satisfy the constraints. However for realistically sized networks, combined with restrictive constraints this method becomes unfeasible because of CPU time restrictions. A new method is proposed that finds the feasible shortest paths directly and can be applied in combination with a wide class of constraints. The paper explains how this CKSP algorithm can be implemented using the ordinary shortest path computation as its elementary operation. An example is provided in which the method is used to enumerate paths while avoiding strongly overlapping and overly circuitous paths. In this context the computational performance of the CKSP method is compared with that of the conventional method. On a network consisting of 200 nodes a speed-up factor exceeding 62 has been demonstrated on a problem that involves finding the 200 constrained shortest paths. The speed-up factor increases sharply with the size of the network and the level of restriction of the constraints. As opposed to the conventional method, the proposed implementation of the CKSP method displays only a limited sensitivity to the level of restriction of the constraints. While the conventional method could only deal with small networks, the proposed method can also enumerate paths for more realistically sized networks.
Conference Paper
In this paper we propose a technique for determining the set of pareto optimal paths and associated probability distributions, as well as the minimum path length distribution for all nodes to a given destination for a directed, cyclic or acyclic network where the arc lengths are given by independent, discrete random variables, whose distributions vary with time and are known. We suggest several approaches for selecting one path from the set of pareto optimal paths determined by the algorithm and discuss some heuristic procedures for further eliminating some paths from consideration
Modeling inter urban route choice 22 behavior
  • M Ben-Akiva
  • M Bergman
  • A Daly
  • R Ramaswamy
Ben-Akiva, M., Bergman, M., Daly, A. and R. Ramaswamy, Modeling inter urban route choice 22 behavior, Ninth International Symposium on Transportation and Traffic Theory, VNU Science 23 Press, Utrecht, Netherlands, 1984.
Multidimensional path search and assignment
  • T De La Barra
  • B Perez
  • J Anez
de la Barra, T., Perez, B. and J. Anez, Multidimensional path search and assignment, Proceedings 27 of the 21st PTRC Summer Meeting, Manchester, UK, 1993.
  • N J Van Der Zijpp
  • S Fiorenzo-Catalano
van der Zijpp, N.J. and S. Fiorenzo-Catalano, Path enumeration by finding the constrained k-1 shortest paths, Transportation Research Part B: Methodological, 39 (6) 545-563, 2005.
Nutzung von Mobilfunkdaten für die Analyse der Routenwahl, PhD Dissertation at 7 the Department of Traffic Engineering and Transportation Planning
  • J Schlaich
8. Schlaich, J., Nutzung von Mobilfunkdaten für die Analyse der Routenwahl, PhD Dissertation at 7 the Department of Traffic Engineering and Transportation Planning, University of Stuttgart 8 Stuttgart, Germany, 2009.