Metodologia de análise de regressão geograficamente ponderada aplicada ao fenômeno das viagens intermunicipais

Abstract

Simple and multiple linear regression models have been largely used in transportation planning studies, specially aiming at forecasting demand during the trip generation stage. However, when there is spatial autocorrelation for the dependent variable, or even for the explicative variables, the analysis of transportation phenomenon requires the consideration of the spatial attribute. This paper has, therefore, the objective of describing a methodology for calibrating geographically weighted linear regression models, presenting as a sample application the analysis of radial trip generation in the intercity public transit system of the State of Ceara.