multi-state non-homogeneous semi-markov model of daily activity type, timing and duration sequence

Tai-Yu Ma (1) , Charles Raux (1) , Eric Cornelis (2) , Iragaël Joly (3)

1 LET - Laboratoire d'économie des transports
2 Groupe de recherche sur les transports
3 GAEL - Laboratoire d'Economie Appliquée = Grenoble Applied Economics Laboratory

Understanding travelers' daily travel-activity pattern formation is an important issue for activity-based travel demand analysis. The activity pattern formation concerns not only complex interrelations between household members and individual's socio-demographic characteristics but also urban form and transport system settings. To investigate the effects of these attributes and the interrelationship between conducted activities, a multistate semi-Markov model is applied. The underlying assumption of the proposed model states that the state transition probability depends on its adjoining states. Based on the statistical tests of significance, it is affirmed that the duration of activity depends not only on its beginning time-of-day but also on the duration of travel/activity previously conducted. The empirical study based on Belgian Mobility survey is conducted to estimate individual's daily activity durations of different episodes and provides useful insight for the effects of socio-demographic characteristics, urban and transportation system settings on the activity pattern formation.

Mots clés: travel-activity pattern; semi-Markov process; Cox regression model

Domaines: Economies et finances

Tai-Yu Ma, Charles Raux, Eric Cornelis, Iragaël Joly. multi-state non-homogeneous semi-markov model of daily activity type, timing and duration sequence. Transportation Research Record, 2009, NO. 2134, pp. 123-134. ⟨10.3141/2134-15⟩. ⟨halshs-00310900v2⟩