Aggregation Issues in Activity-Based Travel Demand Models: Exploration, Understanding and Solutions

University Of Louisville Research Foundation Inc, Louisville KY

Investigators

Abstract

The Urban Transportation Modeling System (UTMS) remains the dominant approach to urban travel demand modeling within the transportation planning profession despite concerns that have been raised regarding its ability to meet the challenges of contemporary and future transportation planning. These concerns have not addressed through the development of new forms of activity-based travel demand models, which have the potential to remedy the deficiencies of UTMS by providing transportation planners with policy-sensitive forecasts. The reliability of these forecasts depends on sound model development, however, and many issues have yet to be addressed on this front. This project seeks answers to two questions relating to the development of activity-based travel demand models. First, what spatial resolution is appropriate for such modeling? Second, what activities are to be modeled? To address these questions, a systematic research program will be undertaken with three objectives: (1) to evaluate the effects of the modifiable areal unit problem (MAUP) on spatial choice sets, consisting of urban opportunities (i.e. potential destinations), and on the parameters of and outcomes from spatial choice models; (2) to develop an optimal zoning system that minimizes MAUP effects in activity-based travel demand models; and (3) to evaluate the effects of aggregating non-work, out-of -home activities on the attributes of those activities and on model parameters estimated for specific attributes. Data will be drawn from two sources. First, information on non-work, out-of -home activities will be obtained from a household travel survey that was conducted in the Louisville, Kentucky, MSA in the fall of 2000. Second, details concerning all urban opportunities in the region will be taken from a database obtained from ReferenceUSA. The geo-processing capabilities of GIS will be used to obtain specific data for many of the empirical analyses. Descriptive statistics and statistical models (i.e. hazard models, multinomial logit models, and ordered probit models) will be used to analyze the data. The priorities of urban transportation planning have changed considerably in recent years, largely due to government legislation such as the Clean Air Act Amendments of 1991. Instead of building roads to meet a growing demand for travel, transportation planning today seeks to better manage existing infrastructure. For this task, policy-sensitive tools like activity-based travel demand models are required. The results of this research will shed considerable light on two issues that have yet to be addressed concerning the development of such models, spatial aggregation and activity aggregation. Potential errors associated with these issues will be reduced, thereby increasing the reliability of forecasts and their usefulness for decision making.