Application of geographically weighted regression to the direct forecasting of transit ridership at station-level

In recent years, station-level ridership forecasting models have been developed based on Geographic Information Systems (GIS) and multiple regression analysis. These models estimate the number of passengers boarding at each station as a function of the station characteristics and the areas that they serve. These models have considerable advantages over the traditional four-step model, including simplicity of use, easy interpretation of results, immediate response and low cost. Nevertheless, the models usually use traditional ordinary least squares (OLS) multiple regression, which assume parametric stability. This study proposes a direct model that uses geographically weighted regression (GWR) to forecast boarding at the Madrid Metro stations. Here, the results obtained using the OLS and GWR models are compared. The GWR model results in a better fit than the traditional one. In addition, the information supplied by the GWR model regarding the spatial variation of elasticities and their statistical significance provides more realistic and useful results.

► Direct demand forecasting (only four explanatory variables) models.
► Calculate of variables based in GIS tools.
► Local fit of regression at station-level.
► GWR method has major performance (lowest residual, not spatial autocorrelated).