This repository provides some of the code implemented and the data used for the work proposed in "A Cluster-Based Trip Prediction Graph Neural Network Model for Bike Sharing Systems".

The datasets used can be found in:

• CitiBike Historical Trips - https://ride.citibikenyc.com/system-data
• Weather - https://mesonet.agron.iastate.edu/request/download.phtml?network=NY_ASOS
• Station Status - https://www.theopenbus.com/raw-data.html
• Geographical Distances - https://openrouteservice.org/

All additional data generated to run the AdaTC clustering algorithm is in the Data folder (the geographical distances extracted from the ORS API are also available). The code is also available in the Code folder.

Contains:

• pairwise_distances.csv.gz - matrix of distances between stations extracted from ORS API according to the regular cycling profile (Note: the stations are indexed in rows and columns by the corresponding citibike id and the matrix is not symmetric - symmetry can be obtained by considering the mean between the two paths directions)
• stations_18.csv.gz - citibike id, name, latitude and longitude of stations in 2018
• df_checkout_pattern_all.csv.gz - dataframe with the tuple of 5 elements corresponding to the check-out patterns in each of the 5 time slots and for each station

The implementation considers the following time slots:

Contains:

• adatc_run.py - code for running adatc algorithm for some intrinsic parameters combination