The First Transportation Engineering Research Symposium
Abstract Details
| Name | School | Title of presentation |
| Rana Al-Jammal | UMass Amherst | Identifying Groups in Data in Search for Unique Patterns of Activity Travel Behavior |
| When modeling activity travel patterns of individuals we are faced with the problem of endless possibilities of individual trip decision-making. With multiple constraints influencing decisions of travel and activity participation each day, a person’s daily activity schedule might not be as predictable as it might appear to be at first glance. Given present constraints, the sequence in which a person schedules daily activities such as: work, study, shopping, leisure, maintenance, etc., affects the number of trips made each day and their duration. Transportation planners are thus interested in forecasting this type of travel behavior in the most accurate way possible. Faced with such complexity of daily schedules that might change for the same person from day to day and week to week, the challenge becomes one of finding a method to categorize similar travel behavior among groups of people in the population to help forecast their trip making decisions. This paper proposes first to use the clustering method to help narrow down the endless possibilities into a manageable group that display similar activity patterns, and second to apply space-time prism concepts to help locate unique activity travel patterns. To accomplish this task, data from the 1998 German Mobidrive six- week travel diary survey is selected for its’ comprehensiveness in representing the reality of trip making with all its complexity. The wealth of information will be very useful in observing the main activity travel patterns occurring in this case study. The goal of this research is to improve the practice of travel behavior forecasting. The use of the cluster analysis and space-time prisms methods in this paper comprise the basis for a framework designed for trip chaining activity travel patterns modeling introduced in a prior paper (Al-Jammal and Parkany, 2003). This paper will present the first two phases of this modeling framework and set the tone for the following phases of analysis and modeling activities. This paper focuses on identifying distinct groups of people and their patterns of travel behavior. Identified population sub-groups will require special attention and separate modeling activity, as they might share different needs and characteristics from other groups and thus might be influenced by certain factors of the modeling process in different intensities. Thus attention will be given to the differences between the population groups and what makes them unique. The first section of this paper will present cluster analysis methods and their application to locate cohesive groups of the population displaying similar activity patterns. The use of cluster analysis as an exploration tool will help distinguish groups in the data of this case study. With a variety of clustering types at hand, the investigative process into the data will be conducted using partition clustering to divide the population into a K number of groups by clustering around the means. A variety of key variables will be selected to aid in this process. The second section of this paper will present the concepts of space-time prisms and their use in modeling activity patterns. Modeling activity patterns is a complicated process not for lack of data but for the endless variety of choices that a person has when it comes to spending time and participating in activities. Although there are many constraints that constrict a person’s activity schedule, a person still possesses the freedom to choose from a multiple of activity options that meet his or her needs. As a person moves through time and space he or she will display a pattern of activity travel behavior. The space demarcated by his or her movement is called an action space. The action space thus has its boundaries that limit a person’s reach but still allows the person to move around within this space. Space-time prisms will be constructed for a representative sample to demonstrate the activity travel patterns of the identified population sub-groups. The innovative application of the space-time prism concepts promises great use in the study of activity patterns. Recognition of complicated patterns will be simplified through the visualization process. Not only this process will allow the study of relationships between activities but also aids in the understanding of motivations behind travel decisions. In this study, the chaining of activities will be uncovered for three population sub-groups to reveal their overall patterns of behavior. Results from this paper will be utilized in future research presenting the second part of the framework for modeling trip chaining behavior developed using activity based travel behavior analysis methods. |
Events Page
Visit the New England Section of ITE events page for news about other upcoming events around New England, including a Mass ITE Chapter meeting and Mass Municipal Engineers Association, in March 2005.
[an error occurred while processing this directive]