Bike Rental and Weather Data across Dozens of Cities


Materials on this website are associated to the following paper:


Introduction and Motivation

Effectiveness of an urban bike rental service depends on how well-planned are the layout of stations and the balancing activities which take bikes from full stations to empty stations. Layout and balancing decisions require the analysis and forecast of bike rental demand. To this date we are not aware of any public bike rental datasets which would cover many cities to facilitate largescale analyses of demand. We provide a dataset of bike rental station status logs from 3584 stations in 27 cities of 11 countries, over the period of 7 months. The dataset is accompanied with weather information from the same cities over the same period. We have performed some example analyses to demonstrate that the dataset provides a rich source for many types of analyses, including analyses about the weekly profiles, station dependencies and relationships between weather and demand. We plan to provide updates to this dataset on a regular basis.

Description of Data

With this paper we present a novel public dataset of bike rental and weather information across dozens of cities.

Data collection started on Sept 26, 2014 and has been running continuously since then. The proposed dataset has 7 months of data ending with April 26, 2015 but we plan to provide new versions of the dataset with new information on a regular basis.

In order to minimise the amount of missing data we use 3 independent downloading servers to collect the data. Nevertheless, there are some missing values due to denials of service by the used open data services. In total, there are 0.7% of missing bike data (37 hours with no data from any stations) and 2% of missing weather data (different missing hours in different cities).

In this dataset's Version 1, there are 3584 stations in 27 cities in 11 countries. Most of the stations (2295) are in the 12 cities of France, and among those most (1240) are in Paris. The station sizes are mostly between 10 and 50 with the median at 20 docks per station.

Feature description:

Used Data Sources

The data have been compiled using the following open data services:


Version 1.0: From Sept 26, 2014 to April 26, 2015

Version 2.0: From Sept 26, 2014 to July 1, 2015