University of Twente Student Theses
As of Friday, 8 August 2025, the current Student Theses repository is no longer available for thesis uploads. A new Student Theses repository will be available starting Friday, 15 August 2025.
Enhancing bicylce route choice predictions : A calibration tool for Fietsmonitor
Velden, J.M.T. van der (2025) Enhancing bicylce route choice predictions : A calibration tool for Fietsmonitor.
|
PDF
1MB |
| Abstract: | Around 28% of the total trips made by Dutch citizens are made by bicycle (de Haas and Kolkowski, (2023)). Because of this, bicycle traffic must be taken into consideration during the decision process regarding (bicycle) infrastructure. Witteveen+Bos has developed the Fietsmonitor to assist stakeholders in this process. The Fietsmonitor is a computer tool that makes predictions on the bicycle traffic in a city using OD-matrices and a ’shortest-path’ route choice model. Previous studies, such as Van Nijen et al., (2024), have found that many more decision variables also impact route choice besides travel distance. This knowledge called for an improvement of the Fietsmonitor: a calibration tool that connects the traffic predictions to the observed bicycle traffic in a city. In this thesis, a calibration tool for the Fietsmonitor is developed, while taking into account the uncertainties in the available input data: the OD-matrix and the observed count data. The calibration tool uses a machine learning approach through gradient descent in order to optimize the uncertain parameters. It does this through four different types of calibration: the route choice probabilities, the traffic within each OD-pair, the overall OD-matrix, and the observed count data. The model performs best if the route choice probabilities are initialized as equal over the different routes, the probabilities are normalized through a standard normalization function, and a learning rate of 0.1 is used. A sensitivity analysis has shown that the calibration tool is robust to changes in the input data. In the end, the calibration tool managed to bring a mean absolute error of 610 bicycles down to a mean absolute error of only 64 bicycles in 100 iterations. This is a significantly better prediction of the bicycle traffic compared to the original Fietsmonitor. Thus, the developed calibration tool was found to be a valuable addition to the Fietsmonitor. |
| Item Type: | Essay (Bachelor) |
| Faculty: | ET: Engineering Technology |
| Programme: | Civil Engineering BSc (56952) |
| Link to this item: | https://purl.utwente.nl/essays/107168 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page