Round norms τlp, p ∈ (1,2] and block norms have been utilised for modelling actual distances in transportation networks. A geometric setting will permit the establishment of a relationship between bias of the road network distance and trajectory deviations, which will be used to separate the set of origin–destination pairs into two samples and also to analyse each sample using regression, thus obtaining several types of estimators. What will be demonstrated in this paper is that these functions can be combined through either a weighted sum, or by means of the introduction of the expected distance concept applied to the bias, to obtain distance predicting functions for the region considered.
