Identifying Behavioral Principles underlying Activity Patt..

Capturing behavioral principles within the context of activity-based travel patterns is of vital importance to build adequate transportation planning models. Of course, there is no solitarily model which is perfectly capable of capturing all behavioral patterns but certain techniques are better suited for it than others. In this paper the technique of Bayesian networks is introduced. Bayesian networks are potentially very strong representation techniques since they are capable of capturing the multidimensional nature of complex decisions. Several arguments are presented which clarify why the presented approach is particularly well suited to identify behavioral patterns. To this end and as an empirical study, several significant findings which might exert influence on the choice of transport mode choice were extracted from a large number of potential factors, in the context of a large activity diary dataset. Furthermore, the paper shows a detailed sensitivity analysis report which enables a quantitative evaluation. Janssens, Wets, Brijs, Vanhoof and Timmermans 3 INTRODUCTION A premise underlying activity-based transportation research is that activity-travel patterns are realizations of behavioral principles that individuals and households use to organize activities in time and space (1). However, gaining insight into behavioral patterns is no easy task at all, since individuals and households use a combination of scheduling principles and mechanisms to cope with the uncertain environment and the complexity of the decision problem (2,3). Since individuals adjust their behavior through complex decision making -motivated by a desire to achieve their own objectives subject to their own personal circumstancesmany institutional approaches turn out to be counter-productive. For instance, many initiatives which intend to restrain car use and improve mobility fail due to a lack of understanding of individual travel behavior. Therefore, transportation planning models which are not capable of capturing such behavioral realism are predetermined to fail (4). Of course, there is no solitarily model which is perfectly capable of capturing all behavioral patterns but certain techniques are better suited for it than others and choosing the appropriate technique may already significantly reduce the bias. The goal of the present paper is to identify factors, which are crucial for individual decision-making and to identify how those factors are interrelated with each other. In particular, this paper deals with the identification and the interpretation of all the factors which might exert influence on transport mode choice in the context of a large activity diary dataset. There are several other research projects, like those by (5) that discuss the problem of identifying transport mode choice by means of artificial intelligence techniques, but Bayesian networks were to the best of our knowledge only used to a very limited extent in the transportation literature (6). In contributing to this line of research, it will be shown that Bayesian networks are potentially very strong representation techniques since they are capable of capturing the multidimensional nature of complex decisions. This can be done by taking many dependencies which exist between the variables into account. Furthermore, the technique is not restricted to the identification of the significant variables but it enables a more quantitative evaluation as well. The paper is organized as follows. First, we will give a brief introduction to Bayesian networks. Next, a description of the data which was used to illustrate the relevance of the approach is given. Furthermore, it is shown that a pruning stage needed to be added to the technique in order to make the application feasible. This is followed by a detailed sensitivity analysis, which is used to identify the differentiators of the transport mode variable and to get a better understanding of the behavioral pattern of individuals for this dataset. BAYESIAN NETWORKS A Bayesian network consists of two components (7): first, a directed acyclic graph (DAG) in which nodes represent stochastic domain variables and directed arcs represent conditional dependencies between the variables and second a probability distribution for each node as represented by conditional dependencies captured with the directed acyclic graph. The representation of a large Bayesian network often results in a tangle of nodes and at first glance this might look confusing but it is above all a very powerful representation tool. In Figure 1, a hypothetical example of a small Bayesian network is depicted. The example describes how different factors may have an impact on the choice of transport mode. Conditional dependencies represented with arcs link a variable, called child variable with the set of its immediate predecessors, called parent variables, according to the direction. For instance, the variable “Driving_license” in Figure 1 is a child variable of the parent variable “Number_cars”. The variables in a Bayesian network can take on values from a limited set of values (states), in which case they are regarded as discrete, or can be continuous and be described with parameters of Gaussian or other distribution for continuous random variables. In Figure 1, each state is shown with its belief level (probability) expressed as a percentage and as a bar graph. Since algorithms, which are used to build Bayesian network structures do not allow the construction of cyclical graphs, some directions are imposed by the algorithm in order to enforce this acyclical relationship. Therefore, the intuitive interpretation of some arrows might look strange and thus it is more appropriate to consider the directed arc as an association rather than as a causality relationship.