“Data-Driven Insights into (E-)Bike-Sharing: Mining a Large-Scale Dataset on Usage and Urban Characteristics - Descriptive Analysis and Performance Modeling” (supplementary dataset)

Scope – This repository accompanies the article in Springer Transportation (DOI 10.1007/s11116-025-10661-2, 2025) and provides the full relational data export (43 Mio km, 2.3 GiB compressed) required to reproduce all analyses, together with 1000‑row excerpts for quick exploration.

Authors

Name	Affiliation	Email	Notes
Felix Waldner*	Technical University of Munich, School of Engineering and Design, Institute of Automotive Technology, Germany	f.waldner@tum.de	*Corresponding author, Equal contribution
Georg Balke	Technical University of Munich, School of Engineering and Design, Institute of Automotive Technology, Germany	georg.balke@tum.de	Equal contribution
Felix Rech	Technical University of Munich, School of Computation, Information and Technology, Germany
Martin Lellep	The University of Edinburgh, School of Physics and Astronomy, United Kingdom

---

Repository layout

.
├── full/                     # full‑resolution CSV exports
│   └── dataset.zip                
│       └── bikes.csv              (≈ 88 k rows )
│       └── bike_types.csv         (≈ 121 rows)
│       └── cities.csv             (267 rows)
│       └── city_areas.csv         (267 rows – WKT/WKB/GeoJSON geometries)
│       └── stations.csv           (≈ 13 k rows)
│       └── station_status.csv     (≈ 38 M rows, 1.7 GiB uncompressed)
│       └── trips.csv              (≈ 25 M rows, 2.7 GiB uncompressed)
├── sample/                   # first 1 000 rows of every file (handy for testing, exploration)
│   ├── bikes.csv
│   ├── …
├── etl/                      # Code used to process the raw data - just informational
│   ├── data_pipeline.py      # Data decompression, explosion, splitting, upload
│   ├── utils.py              # little helper functions for PostgreSQL upload
│   ├── config.template.py    # config file for database credentials and paths
│   └── schema_definition.sql # Data inserts, upserts, trip extraction handled in-database (PostgreSQL)
└── README.md                 # you are here

Data model & file schemas

All timestamps are Unix epoch seconds in UTC; join via cities.timezone to convert to local times. All coordinates are WGS‑84 (EPSG:4326).

trips.csv (25 210 627 rows)

column	unit	description
bike_id	–	Bike identifier; bikes.id
city_id	–	Rental city; cities.id
time_start	s	Rental start (UTC)
lon_start, lat_start, lon_end, lat_end	°	Start & end coordinates WGS-84
station_id_start, station_id_end	–	Station IDs, NULL if free‑floating
battery_start, battery_end	%	State of charge (e‑bikes only)
duration	s	Trip duration
distance	m	Great‑circle distance (PostGIS ST_Distance)

cities.csv (267 rows)

column	unit	description
id	–	Primary key
name	–	Name of the bike‑sharing scheme
lat, lon	°	Approximate city center
timezone	–	IANA timezone string (e.g. Europe/Paris)
country	–	ISO‑3166 alpha‑3 country code
return_to_official_only	bool	true if bikes must be left at a (virtual) station

city_areas.csv (267 rows)

column	description
city_id	cities.id
geom_ewkb	Operational area estimated via DBSCAN in EWKB
geom_ewkt	… in EWKT
geom_geojson	… in GeoJSON

bike_types.csv (121 rows)

column	unit	description
id	–	Technical bike type
vehicle_image	–	URL of representative image
name	–	Commercial name
description	–	Free‑text description
form_factor	–	regular / cargo
rider_capacity	–	Typical seats (1, 2, …)
propulsion_type	–	human / electric
max_range	m	Nominal electric range
battery_capacity	Wh	Battery energy

bikes.csv (88 444 rows)

column	unit	description
id	–	Bike identifier
bike_type_id	–	Technical type; bike_types.id
computer_id	–	On‑board computer identifier

stations.csv (13 192 rows)

column	unit	description
id	–	Station identifier
city_id	–	cities.id
name	–	Human‑readable label
app_number	–	Number shown to users
terminal_type	–	Hardware generation (12 values)
place_type	–	Unknown (23 observed values)
bike_racks	–	Regular parking positions
special_racks	–	Charging racks
lon, lat	°	Location

station_status.csv (38 279 885 rows)

column	unit	description
station_id	–	stations.id
time	s	Snapshot timestamp (UTC)
bikes	–	Total bikes currently docked
booked_bikes	–	Bikes reserved by users
bikes_available_to_rent	–	Bikes ready for rental
free_racks	–	Empty regular docks
free_special_racks	–	Empty charging docks
maintenance	bool	true = offline

---

Quick start

# 1. Clone with Git LFS for large files (optional but recommended)
$ git lfs install
$ git clone https://github.com/tumftm/european-bike‑sharing‑dataset.git

# 2. Decompress the dataset
$ cd full
$ unzip dataset.zip  

# 3. Spin up PostGIS (example)
$ createdb bikesharing
$ psql bikesharing -c 'CREATE EXTENSION postgis;'
$ ogr2ogr -f PostgreSQL PG:"dbname=bikesharing" trips.csv -nln trips -oo X_POSSIBLE_NAMES=lon_start -oo Y_POSSIBLE_NAMES=lat_start

For exploratory work you can start with the 1 000‑row sample files:

import pandas as pd
import geopandas as gpd

trips = pd.read_csv('sample/trips.csv')
# ...
# You might want to perform the EWKB conversion manually
city_areas = gpd.read_file("sample/city_areas.csv")
city_areas = city_areas.set_geometry(gpd.GeoSeries.from_wkb(city_areas.geom_ewkb))

License

The dataset is distributed under the Creative Commons Attribution - NonCommercial 4.0 International (CC BY-NC 4.0) license. When using or adapting the data, please cite the paper which links back to this repository.

@article{waldnerbalke2025,
  title={Data-Driven Insights into (E-)Bike-Sharing: Mining a Large-Scale Dataset on Usage and Urban Characteristics - Descriptive Analysis and Performance Modeling},
  author={Waldner, Felix. and Balke, Georg and Rech, Felix and Lellep, Martin},
  year={2025},
  journal={Transportation}
}

Contact

• Felix Waldner – felix.waldner@tum.de
• Georg Balke – georg.balke@tum.de
• Institute of Automotive Technology, TU Munich, Germany – sekretariat.ftm@ed.tum.de

---