Abstract
This paper presents a sequential method involving application of Principal Component Analysis (PCA) and binomial logit for forecasting of motorized travel mode choice. The application of ACP reduces the multicolinear database into uncorrelated components. Such components are used later as explanatory variables in binomial logit models. The data used are from the Origin-Destination Survey carried out in 2007 in São Paulo Metropolitan Area. The gotten models showed good accuracy and consistent and significant calibrated parameters. In the validation step, the hit rates were obtained ranging from 69% to 92%. Finally, the proposed method is reasonable to be a good alternative for the case of multicollinear data used in regression methods.
Key words:
travel mode choice; Principal Component Analysis; Logistc regression methods
