No abstract

Travel surveys are increasingly taking advantage of GPS data, which offer precise route and time observations and a potentially reduced response burden. In these surveys, travel diaries are usually constructed automatically where research on the employed procedures has been focused on mode identification. The goal of the research reported here was to improve trip purpose identification. The analysis used random forests, a machine-learning approach that had been successfully applied to mode identification. The analysis was based on GPS tracks and accelerometer data collected by 156 participants who took part in a 1-week travel survey in Switzerland that was completed in 2012. The results show that random forests provide robust trip purpose classification. For ensemble runs, the share of correct predictions was between 80% and 85%. Different setups of the classifier were possible and sometimes required by the application context. The training set and its input variables (feature set) of the classifier were defined in various ways. Four relevant setups were tested for this study.

The activity-based multiagent simulation toolkit MATSim adopts a coevolutionary approach to capturing the patterns of people's activity scheduling and participation behavior at a high level of detail. Until now, the search space of the MATSim system was formed by every agent's route and time choice. This paper focuses on the crucial computational issues that have to be addressed when the system is being extended to include location choice. This results in an enormous search space that would be impossible to explore exhaustively within a reasonable time. With the use of a large-scale scenario, it is shown that the system rapidly converges toward a system's fixed point if the agents’ choices are per iteration confined to local steps. This approach was inspired by local search methods in numerical optimization. The study shows that the approach can be incorporated easily and consistently into MATSim by using Hägerstrand's time–geographic approach. This paper additionally presents a first approach to improving the behavioral realism of the MATSim location choice module. A singly constrained model is created; it introduces competition for slots on the activity infrastructure, where the actual load is coupled with time-dependent capacity restraints for every activity location and is incorporated explicitly into the agent's location choice process. As expected, this constrained model reduces the number of implausibly overcrowded activity locations. To the authors’ knowledge, incorporating competition in the activity infrastructure has received only marginal attention in multiagent simulations to date, and thus, this contribution is also meant to raise the issue by presenting this new model.

This paper reports on the development of an agent-based cruising-for-parking simulation using the cellular automaton approach. The software is ready for application in a realworld scenario and for calibration with empirical data currently surveyed at the authors' institute.

As outlined in Section 1.4 and by Figures 1.1 and 1.4, MATSim is based on a co-evolutionary algorithm: Each individual agent learns by maintaining multiple plans, which are scored by executing them in the mobsim, selected according to the score and sometimes modi ed. In somewhat more detail, the iterative process contains the following elements: mobsim The mobility simulation takes one "selected" plan per agent and executes it in a synthetic reality. This may also be called network loading. scoring The actual performance of the plan in the synthetic reality is taken to compute each executed plan's score. replanning consists of several steps: 1. If an agent has more plans than the maximum number of plans (a con guration parameter), then plans are removed according to a (con gurable) plan selector (choice set reduction, plans removal). 2. For some agents, a plan is copied, modi ed and then selected for the next iteration (choice set extension, innovation). 3. All other agents choose between their plans (choice). An agent's plans in a given iteration may be considered the agent's choice set in that iteration. As a result, steps 1 and 2 of replanning modify the choice set, while step 3 implements the actual choice between options. Choice is typically based on the score; higher score plans are more likely to be selected. This is discussed in more detail in Chapters 47 and 49. For the time being, note that the three steps of replanning must cooperate for the approach to work: the plans removal step should remove "bad" plans, the innovation step should generate "good" plans, and the choice should, ingeneral, How to cite this book chapter: