Application of discrete 3-level nested logit model in travel demand forecasting as an alternative to traditional 4-step model

This paper aims to introduce a new modelling approach that represents departure time, destination and travel mode choice under a unified framework. Through it, it is possible to overcome shortages of the traditional 4-step model associated with the lack of introducing actual travellers' behaviours. This objective can be achieved through adopting discrete 3-level Nested Logit model that represents different potential correlation (cross elasticity) among departure time, destination and travel mode alternatives. The proposed model has been estimated and tested by using discretionary trips' data from Eskisehir city, Turkey. In the light of the estimation results, individuals tend to jointly decide on discretionary travel dimensions rather than separately as assumed by the traditional 4-step model. Moreover, the proposed approach shows more flexibility in considering attributes of alternatives along with characteristics of decision makers. That results in a more behavioural travel demand modelling, more accurate future forecasting and more trusted policy implications. The proposed model represents a more accurate and reliable alternative for the first 3-steps of the traditional 4-step model in small-scale planning issues. Finally, the proposed approach is a significant milestone toward obtaining a consistent, efficient and integrated full-scale behavioural-model that consists of all travel demand dimensions.