ANALYSIS OF PREDICTIVE QUALITIES OF DISAGGREGATE MODAL-CHOICE MODELS

Abstract

The results of a study that examined the predictive accuracy and ability of a set of disaggregate, behavioral demand models of modal choice are presented. Although other issues such as sample size, value of time, demand elasticity, and policy predicitons are discussed, the primary objective was to test the validity of disaggregate logit models in forecasting. The analysis is structured around a carefully designed before-and-after study of individual travel behavior as affected by significant, short-term changes in the transportation system. Various specifications of disaggregate modal-choice models are calibrated by using as input data the actual responses of individuals from the before phase of the travel-behavior surveys. This was followed by a series of prediction and validation phases by using the after data that was generated by changes in the transportation system. Because the actual modal shares are known from the longitudinal data, it is possible to assess accurately the predictive qualities of the calibrated logit models. The results of the empirical analysis indicate that disaggregate models, especially those that include a full range of transportation level-of-service and socieconomic variables, can be used to predict future behavior with acceptable levels of performance. /Author/

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 29 citations exist upstream.