A needs-based approach to activity generation for travel demand analysis/

[摘要] This thesis develops a needs-based framework for behavioral enhancement of conventional activity-based travel demand models. Operational activity-based models specify activity generation models based on empirical considerations which are weakly founded in a behavioral theory. This thesis aims to enhance the specification of the activity generation models by developing the conceptual and analytical relationship between individuals;; activity choices and need-satisfaction. The theory of needs hypothesizes that individuals conduct activities to satisfy their needs. Each activity that an individual conducts may satisfy one or several of their needs. Conversely, each need may be satisfied by one or several activities. This thesis models an individual;;s choice of activity dimensions including frequency, sequence, location, mode, time-of-travel, etc. as one that maximizes his/her need-satisfaction. A conceptual model of the relationship between needs and activities is developed based on inventory theory. Every need is associated with a psychological inventory that reflects the level of satisfaction with respect to the need. When an activity that satisfies a need is conducted, the need is satisfied and the corresponding psychological inventory is replenished by a quantity called the activity production. Over time, this inventory gets consumed and the need builds up. The choice of activity dimensions is modeled as a psychological inventory maximizing (i.e. utility-maximizing) problem, subject to time and cost budget constraints. The framework also accounts for satiation in need-satisfaction. An analytical model is formulated, solved and empirically estimated for a single need and the activity that satisfies the need under steady-state conditions. The problem is solved in two stages, for discrete (location) and continuous (duration and frequency) decision variables. The properties of the general solution are studied, and then explored for a translog form of the activity production function. An empirical estimation method that can be applied to single day travel diary data is proposed and validated using Monte-Carlo experiments. The model is empirically estimated using standard travel diary data from the Denver metropolitan area. Estimation results indicate the potential of the needs-based approach to enrich the specification of activity generation models in conventional activity-based model systems. A conceptual framework to extend the single need model is discussed. Extensions to models of multiple needs that capture interactions between different needs are also discussed. The flexible framework can also be extended to model social interactions including intrahousehold activity allocation and joint activity participation by households and social circles. An extension to a dynamic needs-based activity generation model is also discussed, which may be integrated with transportation simulators to predict individuals;; activity choices in response to real-time information.

[发布机构] Massachusetts Institute of Technology