Davy Janssens

Implementation Framework and Development Trajectory of FEATHERS Activity-Based Simulation Platform:

To facilitate the development of dynamic activity-based models for transport demand, the FEATHERS framework was developed. This framework suggests a four-stage development trajectory for a smooth transition from the four-step models toward static activity-based models in the short term and dynamic activity-based models in the long term. The development stages discussed in this paper range from an initial static activity-based model without traffic assignment to a dynamic activity-based model that incorporates rescheduling, learning effects, and traffic routing. To illustrate the FEATHERS framework, work that has been done on the development of static and dynamic activity-based models for Flanders (Belgium) and the Netherlands is discussed. First, the data collection is presented. Next, the four-stage activity-based model development trajectory is discussed in detail. The paper concludes with the presentation of the modular FEATHERS framework, which discusses the functionalities of the modules and how they accommodate the requirements imposed on the framework by each of the four stages.

Integrating Bayesian networks and decision trees in a sequential rule-based transportation model

Several activity-based transportation models are now becoming operational and are entering the stage of application for the modelling of travel demand. Some of these models use decision rules to support its decision-making instead of principles of utility maximization. Decision rules can be derived from different modelling approaches. In a previous study, it was shown that Bayesian networks outperform decision trees and that they are better suited to capture the complexity of the underlying decision-making. However, one of the disadvantages is that Bayesian networks are somewhat limited in terms of interpretation and efficiency when rules are derived from the network, while rules derived from decision trees in general have a simple and direct interpretation. Therefore, in this study, the idea of combining decision trees and Bayesian networks was explored in order to maintain the potential advantages of both techniques. The paper reports the findings of a methodological study that was conducted in the context of Albatross, which is a sequential rule based model of activity scheduling behaviour. To this end, the paper can be situated within the context of a series of previous publications by the authors to improve decision-making in Albatross. The results of this study suggest that integrated Bayesian networks and decision trees can be used for modelling the different choice facets of Albatross with better predictive power than CHAID decision trees. Another conclusion is that there are initial indications that the new way of integrating decision trees and Bayesian networks has produced a decision tree that is structurally more stable.

An integrated activity-based modelling framework to assess vehicle emissions: approach and application

Owing to the richer set of concepts which are involved in activity-based transportation models, the potential advantages of an activity-based approach for air quality purposes have been recognized for a long time. However, models that have been developed along these lines are still scarce. In this research the activity-based model ALBATROSS was used in combination with the emission model MIMOSA to assess the travelled distances and the mobile source emissions produced by passenger cars in the Netherlands. The fact that this approach is based on hourly travel and emission values, rather than on aggregated results or peak hour values, a common practice within other traditional models, is an important added value. The predicted values seem to correspond well with the reported values from the Dutch Scientific Statistical Agency. Predictions for travelled distances overestimated the reported values by approximately 8%. Predictions for emissions of nitrogen oxide, carbon dioxide, volatile organic compounds, and particular matter differed by 16%, 11%, 9%, and 3%, respectively, from the officially reported values. This paper is novel in the sense that it both reports on the applied methodology and presents the practical results from a case study of the activity-based emission modelling approach.

Evaluating the performance of cost-based discretization versus entropy-and error-based discretization

Discretization is defined as the process that divides continuous numeric values into intervals of discrete categorical values. In this article, the concept of cost-based discretization as a pre-processing step to the induction of a classifier is introduced in order to obtain an optimal multi-interval splitting for each numeric attribute. A transparent description of the method and the steps involved in cost-based discretization are given. The aim of this paper is to present this method and to assess the potential benefits of such an approach. Furthermore, its performance against two other well-known methods, i.e. entropy-and pure error-based discretization is examined. To this end, experiments on 14 data sets, taken from the UCI Repository on Machine Learning were carried out. In order to compare the different methods, the area under the Receiver Operating Characteristic (ROC) graph was used and tested on its level of significance. For most data sets the results show that cost-based discretization achieves satisfactory results when compared to entropy-and error-based discretization.Given its importance, many researchers have already contributed to the issue of discretization in the past. However, to the best of our knowledge, no efforts have been made yet to include the concept of misclassification costs to find an optimal multi-split for discretization purposes, prior to induction of the decision tree. For this reason, this new concept is introduced and explored in this article by means of operations research techniques.

Annotating mobile phone location data with activity purposes using machine learning algorithms

Individual human travel patterns captured by mobile phone data have been quantitatively characterized by mathematical models, but the underlying activities which initiate the movement are still in a less-explored stage. As a result of the nature of how activity and related travel decisions are made in daily life, human activity-travel behavior exhibits a high degree of spatial and temporal regularities as well as sequential ordering. In this study, we investigate to what extent the behavioral routines could reveal the activities being performed at mobile phone call locations that are captured when users initiate or receive a voice call or message. Our exploration consists of four steps. First, we define a set of comprehensive temporal variables characterizing each call location. Feature selection techniques are then applied to choose the most effective variables in the second step. Next, a set of state-of-the-art machine learning algorithms including Support Vector Machines, Logistic Regression, Decision Trees and Random Forests are employed to build classification models. Alongside, an ensemble of the results of the above models is also tested. Finally, the inference performance is further enhanced by a post-processing algorithm. Using data collected from natural mobile phone communication patterns of 80 users over a period of more than one year, we evaluated our approach via a set of extensive experiments. Based on the ensemble of the models, we achieved prediction accuracy of 69.7%. Furthermore, using the post processing algorithm, the performance obtained a 7.6% improvement. The experiment results demonstrate the potential to annotate mobile phone locations based on the integration of data mining techniques with the characteristics of underlying activity-travel behavior, contributing towards the semantic comprehension and further application of the massive data.

Improving Performance of Multiagent Rule-Based Model for Activity Pattern Decisions with Bayesian Networks

Several activity-based models are now becoming operational and are entering the stage of application in transport planning. Some of these models use a set of decision trees to support decision making instead of using principles of utility maximization. However, it is believed that the structure of decision trees can sometimes be very unstable and sensitive to highly correlated predictors. Therefore, this study examines whether decision trees constitute the best representational form to capture the behavioral mechanisms and principles that individuals and households use to organize their activities. Findings are reported from experiments conducted by means of Bayesian networks to gain a better understanding of the predictive performance of Albatross, a sequential rule-based model of activity-scheduling behavior. The performances of Bayesian networks and decision trees are compared and results are evaluated by means of detailed quantitative and qualitative analyses. The results showed that Bayesian networks outperformed the decision-tree-based approach for all decision agents of the Albatross model. Given this excellent performance, it is believed that the research community may potentially consider the use of Bayesian networks in developing activity-based transportation models.

The contribution of activity-based transport models to air quality modelling: A validation of the ALBATROSS-AURORA model chain

The potential advantages of using activity-based transport models for air quality purposes have been recognized for a long time but models that have been developed along these lines are still scarce. In this paper we demonstrate that an activity-based model provides useful information for predicting hourly ambient pollutant concentrations. For this purpose, the traffic emissions obtained in a previous application of the activity-based model ALBATROSS were used as input for the AURORA air quality model to predict hourly concentrations of NO(2), PM(10) and O(3) in the Netherlands. Predicted concentrations were compared with measured concentrations at 37 monitoring stations from the Dutch air quality monitoring network. A statistical analysis was performed to evaluate model performance for different pollutants, locations and time periods. Results confirm that modelled and measured concentrations present the same geographical and temporal variation. The overall index of agreement for the prediction of hourly pollutant concentrations amounted to 0.64, 0.75 and 0.57 for NO(2), O(3) and PM(10) respectively. Concerning the predictions for NO2, a major traffic pollutant, a more thorough analysis revealed that the ALBATROSS-AURORA model chain yielded better predictions near traffic locations than near background stations. Further, the model performed better in urban areas, on weekdays and during the day, consistent with the emission results obtained in a previous study. The results in this paper demonstrate the ability of the activity-based model to predict the contribution of traffic sources to local air pollution with sufficient accuracy and confirms the usefulness of activity-based transport models for air quality purposes. The fact that the ALBATROSS-AURORA chain provides reliable pollutant concentrations on hourly basis for the whole Netherlands instead of using only daily averages near traffic stations is a plus for future exposure studies aiming at more realistic exposure analyses and health impact assessments.

Activity-Based Modeling to Predict Spatial and Temporal Power Demand of Electric Vehicles in Flanders, Belgium

Electric power demand for household-generated traffic was estimated as a function of time and space for the region of Flanders, Belgium. An activity-based model was used to predict traffic demand. Electric vehicle (EV) type and charger characteristics were determined on the basis of car ownership and on the assumption that the market shares of EV categories would be similar to the current ones for internal combustion engine vehicles published in government statistics. Charging opportunities at home and work locations were derived from the predicted schedules and the estimation of the possibility to charge at work. Simulations were run for several levels of EV market penetration and for specific ratios of battery-only electric vehicles (BEVs) to pluggable hybrid electric vehicles. A single car was used to drive all trips in a daily schedule. Most of the Flemish schedules could be driven entirely by a BEV even after the published range values were reduced to account for range anxiety and for the overestimated ranges resulting from tests in accordance with standards. The current overnight period for low-tariff electricity was found to be sufficiently long to allow for individual cost optimizing while minimizing the peaks for overall power demand.

Determining Electric Vehicle Charging Point Locations Considering Drivers’ Daily Activities

Abstract In this paper the daily temporal and spatial behavior of electric vehicles (EVs) is modelled using an activity-based (ActBM) micro-simulation model for Flanders region (Belgium). Assuming that all EVs are completely charged at the beginning of the day, this mobility model is used to determine the percentage of Flemish vehicles that cannot cover their programmed daily trips and need to be recharged during the day. Assuming a variable electricity price, an optimization algorithm determines when and where EVs can be recharged at minimum cost for their owners. This optimization takes into account the individual mobility constraint for each vehicle, as they can only be charged when the car is stopped and the owner is performing an activity. From this information, the aggregated electric demand for Flanders is obtained, identifying the most overloaded areas at the critical hours. Finally it is also analyzed what activities EV owners are underway during their recharging period. From this analysis, different actions for public charging point deployment in different areas and for different activities are proposed.

A Conceptual Design of an Agent-based Interaction Model for the Carpooling Application

Abstract Carpooling is an emerging alternative transportation mode that is eco-friendly and sustainable as it enables commuters to save time, travel resource, reduce emission and traffic congestion. The procedure of carpooling consists of a number of steps namely (i) create a motive to carpool, (ii) communicate this motive with other interested agents, (iii) negotiate a plan with the interested agents, (iv) execute the agreed plans and (v) provide a feedback to all concerned agents. The state-of-the-art research work on agent-based modeling is limited to a number of technical and empirical studies that are unable to handle the complex agent behavior in terms of coordination, communication and negotiations. In this paper we present a conceptual design of an agent-based model (ABM) for the carpooling application that serves as a proof of concept. Our agent-based model for the carpooling application is a computational model that is used for simulating the interactions of autonomous agents and to analyze the effects of change in factors related to the infrastructure, behavior and cost. In our agent-based carpooling application we use agent profiles and social networks to initiate our agent communication model and then employ a route matching algorithm and a utility function to trigger the negotiation process between agents. We plan to, as a part of the future work, develop a prototype of our agent-based carpooling application on the basis of the work presented in this paper. Furthermore, we also intend to carry out a validation study of our results with real data.