Travel demand modelling based on time use data

Abstract

Several approaches have been proposed to model travel demand on the basis of activity-patterns. There are modelling concepts where travel behaviour is regarded as the result of discrete choices for the type of activity, departure time, destination, and mode. Other researchers emphasise the interdependencies of different activities and the scheduling process. The idea of the modelling approach which is presented in this paper is to estimate the travel demand in large cities with as few parameters as possible. This ensures that the mechanisms of the simulation do not become too complicated. Very complex models tend to hamper the analysis of the impact of the different influences. Furthermore, efficient computation makes the integration within larger modelling frameworks easier, and different scenarios can be compared in reasonable time. The concept is based on a detailed analysis of diary data. The data where collected in a German nationwide time use survey in 1991/1992. The set of 30,700 diaries is classified by a modified cluster algorithm. This modification allows to handle large data sets. It is possible to distinguish clusters with different levels of heterogeneity. The distances between pairs of diaries, which are needed as input for the clustering procedure, are computed using a sequence alignment algorithm. The sequence alignment method is sensitive not only to different activity budgets but also to the time of the day, when the activities are executed. The classification process reveals which activities cause the differences in the time use patterns, and it turns out that the diary classes possess interesting features when cross tabulated with categories of administrative districts. Especially in rural areas there are pronounced preferences for specific time use patterns. The classes with higher shares of diaries that stem from those areas are different with respect to timing rather than total activity budgets. A new time use survey is conducted in 2001 and 2002, and it is planned to apply the analysis procedures to the new results. As such surveys can not be conducted in every planning or research project, a link between activity-pattern and socio-demographic variables is needed in order to take into account the peculiarities of the area under investigation. This link is established by the correlation of diary classes and socio-demographic groups. The formation of the socio-demographic groups shows which variables are the most important for the estimation of the frequency of time use patterns. The simulation computes activity-travel-patterns for the members of a synthetic population, which serve as input for a traffic flow simulation. The travel times are fed back to the module for the activity-travel-pattern generation and influence the choice of locations, modes and departure times. The simulation model is applied to the city of Cologne in Germany which has about one million inhabitants. The city is the test bed for a conjoint traffic research project, and therefore a good data basis for the generation of a synthetic population is available.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 5 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.