Detecting outlying demand in multi-leg bookings for transportation networks

Abstract

Network effects complicate demand forecasting in general, and outlier detection in particular. For example, in transportation networks, sudden increases in demand for a specific destination will not only affect the legs arriving at that destination, but also connected legs nearby in the network. Network effects are particularly relevant when transport service providers, such as railway or coach companies, offer many multi-leg itineraries. In this paper, we present a novel method for generating automated outlier alerts, to support analysts in adjusting demand forecasts accordingly for reliable planning. To create such alerts, we propose a two-step method for detecting outlying demand from transportation network bookings. The first step clusters network legs to appropriately partition and pool booking patterns. The second step identifies outliers within each cluster to create a ranked alert list of affected legs. We show that this method outperforms analyses that independently consider each leg in a network, especially in highly-connected networks where most passengers book multi-leg itineraries. We illustrate the applicability on empirical data obtained from Deutsche Bahn and with a detailed simulation study. The latter demonstrates the robustness of the approach and quantifies the potential revenue benefits of adjusting for outlying demand in networks.