Trip end models of local rail demand in England and Wales

This paper details models which have been developed to forecast the total number of trips made from local rail stations in England and Wales over a one year period. The use of multiple linear regression and geographically weighted regression in calibration are compared, with both explaining over 75% of the variation in the observed data. The latter technique has not previously been used in rail demand modelling, and allows significant spatial variations in the effect of independent variables to be identified and mapped. A number of catchment definition methods are investigated, as is the inclusion of a wide range of demographic and service related explanatory variables. The models developed are used to forecast usage at stations on the recently opened Ebbw Vale branch line in South Wales and these predictions are compared to initial usage figures.