CC BY 4.0Trautwein, IngoIngoTrautweinEris, M.M.ErisRavlija, DamirDamirRavlijaSonntag, MirkoMirkoSonntag2025-11-172025-11-172025-09-19https://publica.fraunhofer.de/handle/publica/499422https://doi.org/10.24406/publica-643510.1186/s43067-025-00271-410.24406/publica-6435Free-floating sharing with e-mopeds and e-scooters is an upcoming mobility trend. The main problem in shared micromobility is uneven fleet distribution. One option to correct this asymmetry is user-based redistribution. This is especially challenging in free-floating, as the business area must be subdivided. While much research exists for car sharing and station-based bike sharing, the formation and evaluation of sub-areas in free-floating micromobility has barely been analyzed. In this work, we investigate the generation of sub-areas by irregular forms using the Voronoi diagram. The booking data of a free-floating e-moped sharing service in Stuttgart, Germany, serve as basis. Four different generation methods are compared: k-means, density-based spatial clustering of applications with noise (DBSCAN), grid-based clustering, and the use of the underlying road network. The results of the tessellation are evaluated on the basis of geometric properties, data distribution, and adaptability. The k-means algorithm efficiently generates cells in a uniform way, while DBSCAN can adapt well to different demand patterns based on density distribution. Grid-based clustering allows for the consideration of the radius. Metrics for the clusters and polygons were used to compare the methods. It can be seen that k-means (k = 122) performs better in both comparisons. Grid-based ended up in the middle and delivers polygons similar in shape to k-means, while DBSCAN had difficulty adapting to the data. The polygons from the street network generate many polygons, the number and characteristics of which are difficult to influence compared to the other methods.enMicromobilitySharingFree-floatingClusteringTessellationVoronoijournal article