Comparing urban activity travel behaviour

Abstract

An application of activity-based approach to urban travel demand is presented. A day- schedule framework is developed both for workers and non-workers and their travel behaviour is extensively compared. The methodology is applied to two European data sets: the Belgium National Travel Survey and the travel diary survey conducted in two cities of Germany: Halle/Saale and Karlsruhe. A detailed empirical analysis gives insights about the transferability of complex activity based models from a country to another. In the paper both spatial and temporal aspects of mobility are presented: number of tour per day, number of stops per tour, tour and stop activities, interaction between stop making, mode choice, in network time of day for different patterns, distance from stop to stop and activity duration. Two mode choice models for the main activity are estimated both for workers and non- workers. Given the fact that the German survey is a multi-day travel diary a mixed logit formulation has been applied, which allows efficient estimation when there are repeated choices by the same individual. Results show that coefficients in logit with repeated observations are very similar to those estimated in standard logit model, but standard error is much smaller for workers and slightly smaller for non-workers. This observation confirms the fact that workers’ behaviour is homogeneous across weekdays and that non-workers travel behaviour vary a lot across different days. The estimated standard deviations of time are highly significant both for workers and non-workers, indicating that time do indeed vary in the population. These results are encouraging us to pursue the transferability of activity based model to national travel survey data and to apply mixed logit formulation to model most of the components of the activity-travel framework.