Estimating Freight Traffic of Competing Transportation Modes: An Application of the Linear Discriminant Function.

The paper examines the feasibility of using discriminant analysis for estimating allocation of freight traffic among different modes of transportation. The discriminant analysis, although originally developed for problems of other sciences, is finding increasing applications in economics. In transport economics, the discriminant analysis technique has been used for predicting the choice of travel mode for the journey to work, and it appears to be equally useful in analyzing freight traffic. The data collected in the 1963 U.S. Census of Transportation provided an opportunity to apply the linear discriminant function to a sample of shipments. The problems of statistical analysis for which the discriminant functions have been used involve differentiating between members of two or more populations by means of a set of measurements. The data requirement will probably be the major limiting factor in the use of discriminant functions for estimation. Although the discriminant functions for various commodities could be calculated for the data collected by the Census of Transportation, the estimation would require detailed information, on individual shipments.