Tim Hilgert

Modeling Week Activity Schedules for Travel Demand Models

Activity schedules are an important input for travel demand models. This paper presents a model to generate activity schedules for one week. The approach, called actiTopp, is based on the concept of utility-based regression models and stepwise modeling. In contrast to most of the existing models, actiTopp covers the time period of one week. Few models have covered one week; thus, the activity generation approach of this simulation period is rare. Analysis of weekly activity behavior shows stability between different days (e.g., working durations). Hence, the model explicitly takes these aspects into account, for example, by defining time budgets to spread durations within the week. For model estimation, the study used data from the German Mobility Panel (MOP). This annual survey collects representative data on the travel behavior of the German population. The data from 2004–2013 provide more than 17,500 activity schedules for one week, with more than 450,000 activities. Selected results are shown for the model application to 2014 MOP data, which the study used for validation purposes. The mean value of activities per person and week show a difference of 0.3 activity. To evaluate the model, the study used Kolmogorov-Smirnov tests with a significance level of α = 0.001. For the activity type distribution of the 2014 sample, the analysis could not reject the null hypothesis of equality of the distribution of the model and the survey data at this significance level.

Large-Scale Application of a Combined Destination and Mode Choice Model Estimated with Mixed Stated and Revealed Preference Data

The diffusion of new modes of transportation, such as carsharing and electric vehicles, makes it necessary to consider them along with traditional modes in travel demand modeling. However, there are two main challenges for transportation modelers. First, the new modes’ low share of usage leads to a lack of reliable revealed preference data for model estimation. Stated preference survey data are a promising and well-established approach to close this gap. Second, the state-of-the-art model approaches are sometimes stretched to their limits in large-scale applications. This research developed a combined destination and mode choice model to consider these new modes in the agent-based travel demand model mobiTopp. Mixed revealed and stated preference data were used, and new modes (carsharing, bikesharing, and electric bicycles) were added to the mode choice set. This paper presents both challenges of the modeling process, mainly caused by large-scale application, and the results of the new combined model, which are as good as those of the former sequential model although it also takes the new modes into consideration.

Are Activity Patterns Stable or Variable? Analysis of Three-Year Panel Data

Routines and mandatory activities, such as work and school, shape the activity patterns of individuals and strongly influence travel demand. Knowledge about stability and variability of these routines could strengthen travel demand modelling and forecasting. A longitudinal perspective is required to investigate these aspects. In this study, the activity patterns of a sample of people is compared for one week in two successive years. It is analyzed whether the activity patterns of a given person vary from year to year, to what degree, and how this variability and stability can be measured. It is considered whether socio-demographic factors and life events determine stability in weekly activity patterns. The study is based on the representative panel survey, German Mobility Panel. The weekly activity patterns of the same respondents in different years is assessed, using two methods to measure stability and variability. The survey respondents are clustered into three groups according to the degree of variability in their activity patterns. A logistic regression model is also used to identify socio-economic and demographic covariates for similarity in weekly activity patterns. Results show that about one-third of the sample had the same or very similar weekly activity patterns in the two years examined. A person’s occupation status is a good predictor for the variability of activity patterns. Moreover, persons undergoing a change in occupation status are quite likely to show a greater variability in their activity patterns.