DuEPublico 2

Dies ist unser neues Repositorium, derzeit für E-Dissertationen und ausgewählte weitere Publikationen. Weitere Informationen...

Analyse, Modellierung und Simulationen von Routenwahlverhalten

Chmura, Thorsten

This thesis will give an overview about the connection of physics and economy in special view of human behavior in traffic scenarios. Main topic is the human decision making process in a two-route-choice scenario. In three main Experiments, human players had to decide between two routes which connect two cities. The first experiment is defined by a pure Nash-equilibrium and 18 participants, the second experiment is similar, but alternating construction sites on both routes were added. Experiment 1 and 2 are played over 200 periods. The third experiment is a minority game with mixed equilibria and 9 participants. Experiment 3 was played over 100 periods. Furthermore the first experiment was extended to a larger group of players up to 90 persons. For every experiment two treatments with 6 observations were played. The treatments differ in the given information to the participants before they choose between the two routes. In the first treatment the information about the own travel time was given, additionally in the second treatment the information about travel times on both routes were given. The second part of the thesis describes the statistical data and illustrates the results. Two main results are that the equilibrium was not stable in the experiments, the fluctuations persists to the end of every observation. Further on two player types were found and explained in this chapter. In the third part of this thesis an extended payoff-sum model is described and explained. Simulations using this reinforcement learning model were run and show the reproducibility of the empirical data. In the conclusions the main results are shown and discussed.

Teilen und Zitieren


Chmura, Thorsten: Analyse, Modellierung und Simulationen von Routenwahlverhalten.


Nutzung und Vervielfältigung:
Alle Rechte vorbehalten