Alternative Modelling to Estimate Freight Trip Generation

Abstract

Freight trip generation models essentially use the least-squares estimation technique, losing in part the power of weighted least squares, least squares techniques in two stages, and instrumental variables that could address the complexities of the transport phenomenon that has repercussions in the explanatory and dependent variables, as well as in the hypotheses related to the behaviour of the estimation error. In this sense, this paper presents an alternative modelling to estimate freight trip generation using different techniques dealing with the problems arising from linear regression and obtaining efficiency in the estimation models. Also, an exercise considering warehouse freight trip generation modelling is presented using the traditional technique of least squares and comparing the results with least-squares weighted, least-squares techniques in two stages and instrumental variables. The results indicate that the proposed modelling improve linear models to predict freight trip generation estimation, guaranteeing better adherence than traditional modelling.