One of the important problems studied in the area of travel behavior analysis is travel mode choice which is one of the four crucial steps in transportation demand estimation for urban planning. State of the art models in travel demand modelling can be classified as trip based; tour based; and activity based. In trip based approach, each individual trips is modelled as independent and isolated trips i.e. no connections between different trips. In tour based approach, trips that start and end from the same location (home, work, etc) and trips within a tour are dependent on each other. In past two decades, researchers have focussed on activity based modelling, where travel demand is derived from the activities that individuals need/wish to perform. In this approach, spatial, temporal, transportation and interpersonal interdependencies (in a household) constrain activity/travel behaviour.
This paper extends tour-based mode choice model, which mainly includes individual trip level interactions,
to include linked travel modes of consecutive trips of an individual. Travel modes of consecutive trip made
by an individual in a household have strong dependency or co-relation because individuals try to maintain
their travel modes or use a few combinations of modes for current and subsequent trips. Traditionally, tour
based mode choice models involved nested logit models derived from expert knowledge. There are
limitations associated with this approach. Logit models assumes i) specific model structure (linear utility
model) in advance; and, ii) it holds across an entire historical observations. These assumptions about the
predefined model may be representative of reality, however these rules or heuristics for tour based mode
choice should ideally be derived from the survey data rather than based on expert knowledge/judgment.
Therefore, in this paper, we propose a novel data-driven methodology to address the issues identified in tour
based mode choice. The proposed methodology is tested using the Household Travel Survey (HTS) data of
Sydney metropolitan area and its performances are compared with the state-of-the-art approaches in this area.