Chapter 1 - An Introduction to Disaggregate Demand Modeling in the Transportation Field

Abstract

This chapter discusses disaggregate demand modeling in the transportation field. In specifying a model, one must decide the attributes that are going to be used, the number of parameters that one is to estimate, and the functional way in which attributes and parameters are combined in the choice function. The following four features should be carefully considered when selecting attributes: (1) availability, (2) statistical fit, (3) reasonableness, and (4) relevance. An attribute should be available as a part of the data in both the calibration and the prediction stage as otherwise such stages cannot be carried out. It should be noted, though, that an attribute that is not forecastable for certain problems may be forecastable in other instances. The reasonableness is a variable that should be considered for inclusion in the model only if there is some strong a priori feeling that there is some cause-and-effect relationship between such variable and the choice probability. The statistical fit feature includes the standard statistical properties that one likes to see after the model is calibrated. The relevance suggests a reason for including an attribute in the model. As the purpose of demand modeling is to assess the consequences (effects) of certain actions (causes), it seems desirable to include in the model variables that describe the actions.