A Utility-Theory Consistent Model of Activity Frequency, Purpose, and Duration Choices
Corresponding Author: Rajesh Paleti, Old Dominion University
Presented By: Rajesh Paleti, Old Dominion University
Abstract
Activity participation and time-use choices constitute the core of activity-based travel demand models. On any given day, each person can choose to participate in multiple out-of-home activities of different purposes and for different duration. So, there are three dimensions to the activity-travel choices that people make at the planning stage – out-of-home activity participation frequency, activity purpose of the activity episodes, and duration associated with each episode. There are more choices including location and time-of-day choices associated with each activity that are made at the scheduling stage.
The objective of the proposed study is to model the three choices in the planning stage, namely activity frequency, activity purpose, and activity duration, in an integrated manner. This is important because is very likely that these choices are inter-related in nature. For instance, a person can choose to participate in a single out-of-home activity but for longer duration or more activities of different purposes but for shorter duration in each of the activities. Such interactions among these three activity-travel choices can be captured better in a joint model as opposed to models that operate sequentially or independently. It is also important that the integrated modeling framework is easy to implement and is consistent with the well-established behavioral theories in choice literature. However, two of the three dimensions considered in this study, namely activity frequency and duration, are count and continuous response variables. Typically, models used to model such outcomes (e.g., ordered response and hazard duration/linear regression models) are not consistent with the random utility maximization (RUM) theory that forms the basis for most of the discrete choice models used in transportation literature.
The proposed research will develop models that are practically easier to implement and are also consistent with the RUM theory. The models will be estimated using the National Household Travel Survey (NHTS) data for the state of Virginia. Post-estimation analysis including statistical fit comparison and elasticity analysis will be undertaken to demonstrate the applicability of the models developed.