Incorporating the habit effect into scoring function of agent-based transport simulation

Abstract

In studies of travel behavior using small-scale stated preference (SP) survey data, mode choice models have been used to predict the effects of modal shifts when the travel time or cost of a target transport mode changes. Recently, however, some studies have shown the existence of state dependence (also referred to as inertia or habit) in transport mode use, suggesting that failing to account properly for the effects of state dependence may lead to overestimated predictions of the impact of transport policies. State dependence refers to the inertia of travelers who continue using the same mode of transport even when their residential locations or the levels of transport service change. By contrast, large-scale origin-destination (OD)-based transport demand models (such as agent-based models [ABMs]) have not identified the need to control for state dependence, nor have there been related studies on this issue. This gap is due to the difficulty of implementing state-dependent models within OD-based or ABMs, whereas state-dependent mode choice models can be developed using small-scale SP data. Additionally, there is concern that failing to control for state dependence in ABMs may lead to an overestimation of the effects of transport policies. This study proposes a method to incorporate the habit effect into the scoring function of ABMs and implements the related code. Using a public transportation enhancement policy scenario, the study demonstrates that the habit effect influences the results.