Transfer learning approach to bicycle-sharing systems' station location planning using OpenStreetMap data

Raczycki, Kamil; Szymański, Piotr

doi:10.1145/3486626.3493434

Cited by 3 publications

(1 citation statement)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…2 of 18 and give every obtained grid a feature vector that can be correlated with the trajectory data within a machine learning process [9][10][11]. Many methods [2,[12][13][14][15] exceed in predicting large-scale person movement, e.g., to assess city planning regarding transportation capabilities. Others concentrate on individual POI-check-in synthetization.…”

Section: Introductionmentioning

confidence: 99%

Privacy Preserving Human Mobility Generation using Grid based Data and Graph Autoencoders

Netzler,

Lienkamp

2024

Preprint

View full text Add to dashboard Cite

The proposed method deals with the problem of data privacy and sharing when processing personal mobility tracking data. Previous methods concentrate on producing highly detailed data on short-term and restricted areas, e.g. for autonomous driving scenarios. Another possibility consists of city-wide scales and beyond, that are used to predict general traffic flows. The presented approach takes the tracked mobility behavior of individuals to create coherent new mobility data that reflects the long-term mobility behavior of the person, guaranteeing location persistency and sound embedding within the point-of-interest structure of the observed area. After an analysis and clustering step with the original data, the area is distributed into a geospatial grid structure (H3 is used here), and the neighborhood relationships between the grids are interpreted as a graph. A feed-forward-autoencoder and a graph encoding-decoding network generate a latent space representation of the area. The original clustered data is associated with their respective H3 grids. With a greedy algorithm approach and concerning privacy strategies, new combinations of grids are top-level patterns for individual mobility behavior. Concrete locations are found and connected within the grids based on the original data. The described method is then applied to a study with 1000 participants from the city of Munich in Germany, and the results are described, showing the application of the approach in generating synthetic data, enabling further research on individual mobility behavior and patterns. The result is a sharable dataset on the same abstraction level as the input data, which makes it interesting, particularly for machine learning applications.

show abstract