Semantic Annotation of GPS Traces: Activity Type Inference

in Transportation Research Record: Journal of the Transportation Research Board (in press)

Due to the rapid development of technology, larger data sets concerning activity travel behavior become available. These data sets often lack semantic interpretation. This implies that annotation in terms ... [more ▼]

Due to the rapid development of technology, larger data sets concerning activity travel behavior become available. These data sets often lack semantic interpretation. This implies that annotation in terms of activity type and transportation mode is necessary. This paper aims to infer activity types from GPS traces by developing a decision tree-based model. The model only considers activity start times and activity durations. Based on the decision tree classification, a probability distribution and a point prediction model were constructed. The probability matrix describes the probability of each activity type for each class (i.e. combination of activity start time and activity duration). In each class, the point prediction model selects the activity type that has the highest probability. Two types of data were collected in 2006 and 2007 in Flanders, Belgium, i.e. activity travel data and GPS data. The optimal classification tree constructed comprises 18 leaves. Consequently, 18 if-then rules were derived. An accuracy of 74% was achieved when training the tree. The accuracy of the model for the validation set, i.e. 72.5%, shows that overfitting is minimal. When applying the model to the test set, the accuracy was almost 76%. The models indicate the importance of time information in the semantic enrichment process. This study contributes to future data collection in that it enables researchers to directly infer activity types from activity start time and duration information obtained from GPS data. Because no location information is needed, this research can be easily and readily implemented to millions of individual agents. [less ▲]

Semantic Annotation of GPS Traces: Activity Type Inference

in Proceedings of the 92nd Annual Meeting of the Transportation Research Board (DVD-ROM) (2013)

Due to the rapid development of technology, larger data sets concerning activity travel behavior become available. These data sets often lack semantic interpretation. This implies that annotation in terms ... [more ▼]

Due to the rapid development of technology, larger data sets concerning activity travel behavior become available. These data sets often lack semantic interpretation. This implies that annotation in terms of activity type and transportation mode is necessary. This paper aims to infer activity types from GPS traces by developing a decision tree-based model. The model only considers activity start times and activity durations. Based on the decision tree classification, a probability distribution and a point prediction model were constructed. The probability matrix describes the probability of each activity type for each class (i.e. combination of activity start time and activity duration). In each class, the point prediction model selects the activity type that has the highest probability. Two types of data were collected in 2006 and 2007 in Flanders, Belgium, i.e. activity travel data and GPS data. The optimal classification tree constructed comprises 18 leaves. Consequently, 18 if-then rules were derived. An accuracy of 74% was achieved when training the tree. The accuracy of the model for the validation set, i.e. 72.5%, shows that overfitting is minimal. When applying the model to the test set, the accuracy was almost 76%. The models indicate the importance of time information in the semantic enrichment process. This study contributes to future data collection in that it enables researchers to directly infer activity types from activity start time and duration information obtained from GPS data. Because no location information is needed, this research can be easily and readily implemented to millions of individual agents. [less ▲]

Knowledge of the Concept Light Rail Transit: Determinants of the Cognitive Mismatch between Actual and Perceived Knowledge

in Proceedings of the 13th International Conference on Travel Behaviour Research (2012)

The Flemish public transport company “De Lijn” is planning the development of a new Light Rail network for medium range distance trips (10 to 40km). A challenge exists in the fact that the concept of ... [more ▼]

The Flemish public transport company “De Lijn” is planning the development of a new Light Rail network for medium range distance trips (10 to 40km). A challenge exists in the fact that the concept of Light Rail Transit (LRT) is relatively unknown in Flanders. Therefore this paper explores the knowledge of the concept ‘Light Rail Transit’ among the Flemish population. To investigate the knowledge, two separate binary logit models are estimated to explore the determinants of the overall actual knowledge and the determinants of a cognitive mismatch. The results show that age, sex, public transit use, household size, bicycle ownership and weekly number of shopping activities contribute significantly to the overall actual knowledge of the LRT-concept. Besides, cognitive mismatch is only significantly affected by age and gender. Moreover, the results reveal a serious lack of knowledge of the concept of LRT. Consequently, a successful implementation of the LRT-system in Flanders may be jeopardized and thus it is of crucial importance to raise the level of knowledge. A first option is knowledge acquisition based on experience of the transit network. In this view, it can be a good idea to develop “travel-one-day-for-free” marketing actions. Second, it is important to provide information to the travelers by contriving information campaigns based on the determinants identified by the models. How the campaigns should be constructed from an intrinsic and psychological point of view and deliberating between the methods of communication to reach the various target groups are some important considerations for further research. [less ▲]

Identifying the Determinants of Light Rail Mode Choice for Medium/Long Distance Trips: Results from a Stated Preference Study

in Proceedings of the 91st Annual Meeting of the Transportation Research Board (DVD-ROM) (2012)

The introduction of new public transport systems can influence society in a multitude of ways, ranging from modal choices and the environment to economic growth. This paper examines the determinants of ... [more ▼]

The introduction of new public transport systems can influence society in a multitude of ways, ranging from modal choices and the environment to economic growth. This paper examines the determinants of light rail mode choice for medium/long distance trips (10-40km) for a new light rail system in Flanders, Belgium. To investigate these choices, the effects of various transport system specific factors (i.e. travel cost, in-vehicle travel time, transit punctuality, waiting time, access/egress time, transfers, and the availability of empty seats) as well as the travelers’ personal traits, are analyzed using an alternating logistic regression model, which explicitly takes into account the correlated responses for binary data. The data used for the analysis stem from a stated preference survey which was conducted in Flanders, Belgium. The modeling results yield findings that are in line with literature: most transport system specific factors as well as socio-economic variables, attitudinal factors, perceptions and the frequency of using public transport contribute significantly to the preference of light rail transit. In particular, it is shown that the use of light rail is strongly influenced by travel cost and in-vehicle travel time and to a lesser extent by waiting and access/egress time. It also appeared that seat availability plays a more important role than transfers in the decision process to choose light rail transit. The findings of this paper can be used by policy makers as a frame of reference to make light rail transit more successful. [less ▲]

Optimizing the implementation of policy measures through social acceptance segmentation

in Transport Policy (2012), 22

This paper proposes Q-methodology as a technique for the identification of more homogeneous subgroups or ‘segments’ within a rather heterogeneous overall population when it comes to social acceptance of ... [more ▼]

This paper proposes Q-methodology as a technique for the identification of more homogeneous subgroups or ‘segments’ within a rather heterogeneous overall population when it comes to social acceptance of demand-restricting policy measures. Identification of such segments would allow policy makers to better tailor their future actions and thereby increase the chance for a successful implementation of the measures they propose. A set of 33 persons, selected in function of age, gender and car ownership evaluated the acceptability of a total number of 42 demand-restricting policy measures. Special care was taken that the final set of statements covered the four classically distinguished demand-restricting strategies, i.e., improved transport options, incentives for the use of alternative transport modes, parking and land-use management, and institutional policy revision. In addition, a balance between both ‘hard’ and ‘soft’ and ‘push’ and ‘pull’ measures was strived for. The results indicate that four different segments in terms of social acceptance of demand-restricting policy measures can be distinguished, i.e., travelers in favor of traffic calming, travelers against hard push measures, travelers in favor of demand restriction, and travelers against policy innovations. Besides the differences and similarities between these segments, the practical implications for policy makers are discussed, together with a series of specific recommendations and suggestions for future research. [less ▲]

Onderzoek Verplaatsingsgedrag Vlaanderen 4.2 (2009-2010): Analyserapport

Report (2011)

The dual role of weather forecasts on changes in activity-travel behavior

Conference (2011)

In previous studies, conflicting results could be found regarding the impact of weather forecasts on travel decisions, e.g. Khattak and De Palma (1997) found no significant effect of acquiring forecasted ... [more ▼]

In previous studies, conflicting results could be found regarding the impact of weather forecasts on travel decisions, e.g. Khattak and De Palma (1997) found no significant effect of acquiring forecasted weather information on the probability of adapting mode and departure time, whereas the results reported by Hagens (2005), Niina (2009) and Kilpelainen and Summala (2007) indicated that weather forecast do play an important role. Therefore this paper investigates the changes in activity-travel behavior in response to weather forecasts. The data for this study is collected by means of a stated adaptation survey, which is both administered on the internet and via a traditional paper and pencil questionnaire. In total, 595 respondents completed the survey. To obtain an optimal correspondence between the true population and the sample weights are assigned to the observation. Results indicate that weather information plays a dual role. On the one hand people do alter their activity-travel behavior in response to weather information, albeit these changes are not as pronounced when compared to actual weather. On the other hand the extent (frequency and media type) to which people are exposed to these weather forecasts appears to play only a marginal role. This dual role weather information plays in this study appears to be supported by the conflicting international literature and therefore revealing the underlying psychological motivations to change one's activity-travel behavior is a key challenge for further research. [less ▲]

An estimation of total vehicle travel reduction in the case of telecommuting. Detailed analyses using an activity-based modeling approach.

in Proceedings of the 39th European Transport Conference (2011)

Transportation Demand Management (TDM) is often referred to as a strategy adopted by transport planners with the goal to increase transport system efficiency. One of the potential measures that can be ... [more ▼]

Transportation Demand Management (TDM) is often referred to as a strategy adopted by transport planners with the goal to increase transport system efficiency. One of the potential measures that can be adopted in TDM is the implementation of telecommuting. A significant number of studies have been conducted in the past to evaluate the effect of telecommuting on the amount of peak-period trips. However it is less studied whether telecommuting also effectively and significantly reduces total vehicle travel in terms of kilometers traveled throughout the day. For this reason, a conventional modeling approach was adopted in this paper to calculate total kilometers of travel saved in the case telecommuting would materialize in the Flanders area. In a second part, this paper introduces the use of an activity-based modeling approach to evaluate the effect of telecommuting on a more detailed time scale. As the second approach provides a more disaggregate result, both models can be compared on the more aggregate level to validate whether they correspond. [less ▲]

A Bayesian Approach for Modeling Origin-Destination Matrices

in Proceedings of the 90th Annual Meeting of the Transportation Research Board (DVD-ROM) (2011)

The majority of Origin Destination (OD) matrix estimation methods focus on situations where weak or partial information, derived from sample travel surveys, is available. Information derived from travel ... [more ▼]

The majority of Origin Destination (OD) matrix estimation methods focus on situations where weak or partial information, derived from sample travel surveys, is available. Information derived from travel census studies, in contrast, covers the entire population of a specific study area of interest. In such cases where reliable historical data exist, statistical methodology may serve as a flexible alternative to traditional travel demand models by incorporating estimation of trip-generation, trip-attraction and trip-distribution in one model. In this research, a statistical Bayesian approach on OD matrix estimation is presented, where modeling of OD flows, derived from census data, is related only to a set of general explanatory variables. The assumptions of a Poisson model and of a Negative-Binomial model are investigated on a realistic application area concerning the region of Flanders on the level of municipalities. Problems related to the absence of closed-form expressions are bypassed with the use of a Markov Chain Monte Carlo algorithm, known as the Metropolis-Hastings algorithm. Additionally, a strategy is proposed in order to obtain predictions from the hierarchical, Poisson-Gamma structure of the Negative-Binomial model conditional on the posterior expectations of the mixing parameters. In general, Bayesian methodology reduces the overall uncertainty of the estimates by delivering posterior distributions for the parameters of scientific interest as well as predictive distributions for future OD flows. Predictive goodness-of-fit tests suggest a good fit to the data and overall results indicate that the approach is applicable on large networks, with relatively low computational and explanatory data-gathering costs. [less ▲]

Assessment of the Effect of Micro-Simulation Error on Key Travel Indices: Evidence from the Activity-Based Model FEATHERS

in Proceedings of the 90th Annual Meeting of the Transportation Research Board (DVD-ROM) (2011)

Current transportation models often do not explicitly address the degree of uncertainty in travel forecasts. Of particular interest in activity-based travel demand models is the model uncertainty that is ... [more ▼]

Current transportation models often do not explicitly address the degree of uncertainty in travel forecasts. Of particular interest in activity-based travel demand models is the model uncertainty that is caused by the statistical distributions of random components, i.e. micro-simulation error. Therefore, the main objective of this paper is to assess the impact of micro-simulation error on two key travel indices, namely the average daily number of trips per person and the average daily distance traveled per person. The effect of micro-simulation error will be investigated by running the activity-based modeling framework FEATHERS 200 times using the same 10% fraction of the population. Results show that micro-simulation errors are limited especially when disaggregation is limited to two levels. Notwithstanding, results indicate that for more elaborate analyses a 10% fraction might not be sufficient. The size of micro-simulation error increases along with complexity. Moreover, more commonly used transport modes such as using the car as driver have a lower error rate. Further research should investigate the impact of the population fraction on the micro-simulation error rates. Besides, one could also investigate other aspects (e.g. the number of activities) involved in the activity-scheduling process. [less ▲]