Geographic transferability of urban travel generation models: Comparison of regression models and trip rates for some Colombian cities

Abstract

The transferability of models has important practical scope because using parameters of other contexts in order to estimate models in areas with few resources or with very small samples available, constitutes an economic solution to solve planning issues. It is important to analyse the transferability of trip generation models because if we verify the existence of certain patterns of travel, we could save time and money in implementing large mobility surveys. This article examines the geographical transferability of trip generation models based on household surveys in six Colombian cities. Regression models and trip rates were calibrated based on variables such as the rate of motorization, socioeconomic status, and household size. We found that when comparing aggregate models such as linear regression, it is easier to check the transferability of parameters between cities, while in the case of rate models travel the hypothesis of equality is not met ever possibly due to the variation the reliability of the cell values of the model, especially in the extreme categories which have fewer observations.