As of 2025, the Swiss Federal Railways (SBB) use SynPop, a partly synthesized dataset combining demographic, socio-economic, and spatial data from sources like STATPOP and STATENT, for its transport modeling. Developed with ARE, SynPop supports mobility simulations and strategic planning, including the 2050 National Transport Perspectives. In the past, previous efforts to apply k-Anonymity yielded mixed results. While access to these datasets currently requires a form of NDA, the long-term goal is to enable unrestricted public access.

The procedure of the thesis is pretty straightforward. First, like a normal data science pipeline, the original datasets are analysed from a k-Anonymity perspective and discussed, afterwards the anonymization process is being carried out using a framework that was specifically built and coded in R. The results, before and after anonymization, are then evaluated and discussed, leading to conclusions and recommendations from a data anonymization standpoint.

The thesis delivers a functional and adaptable anonymization framework, capable of handling diverse datasets through parameter adjustments. Beyond technical implementation, the in-depth analysis of the original datasets enabled the formulation of specific recommendations, particularly regarding the quality and structure of synthetic variables already present in the data. The study also identifies key limitations and outlines potential next steps for improving data privacy practices for the SynPop dataset. Notably, this thesis represents a first effort in addressing k-Anonymity and target record swapping within SBB’s data ecosystem, laying a foundation for future research and studies on the matter.