Bike Distribution Model for Urban Data Applications

Bike Distribution Model for Urban Data Applications

Maricica Nistor André Dias

CEiiA/Centre of Engineering and Product Development, Portugal

Available online: 
| Citation



Bike sharing systems are fundamental sources of data for creating applications of monitoring the city and guiding the user’s choice for bike usage. Although many related works analyse the generated data by these urban bike systems with the scope of finding bike usage patterns, the variety of the cities and the lack of impact factors require further deeper investigations. We propose a simple, but efficient mathematical approach based on a Markovian model to predict the bike distribution for an urban sharing bike system considering the weather and event impacts. The model is applied for data collected from the New York city bike system. The main findings are relevant for the urban applications and are summarized as follows: (a) the model results substantially address the city’s characteristics, i.e., for the New York city, in terms of weather, only the temperature influences the bike usage, while regarding the events, the impact is insignificant, (b) the hourly bike distribution is predicted 1 day-ahead that is of particular interest to the city manager and (c) to the user who is able to know 1 day in advance the probability of finding an available bike or a free parking space at a specific station. Further city comparison analysis in terms of traffic, vehicle utilization and population density is provided for future purposes. Finding the precise station’s capacity is a forthcoming feature of the proposed model.


data applications, mathematical modelling, urban bike distribution


[1] Sustainable Development Goals, availabe at (accessed August 2018).

[2] Frade, I. & Ribeiro, A., Bicycle sharing systems demand. Procedia – Social and Behavioral Sciences, 111, pp. 518–527, 2014.

[3] Gast, N., Massonnet, G. & Reijsbergen, D., Probabilistic Forecasts of Bike-Sharing Systems for Journey Planning. Proceedings of the 24th ACM International on Conference on Information and Knowledge Management, Melbourne, Australia, 2015.

[4] Singhvi, D., Singhvi, S., Frazier, P., Henderson, S., Mahony, E., Shmoys, D. & Woodard, D., Predicting Bike Usage for New York Cities Bike Sharing System. AAAI Workshop: Computational Sustainability, Texas, USA, 2015.

[5] Lin, L., Peeta, S., He, Z. & Wen, X., Predicting Station-level Hourly Demands in a Large-scale Bike-sharing Network: A Graph Convolutional Neural Network Approach, available at, 2018.

[6] Datta, A., Master thesis, 2014, Predicting bike-share usage patterns with machine learning, University of Oslo.

[7] Yang, Z., Hu, J., Shu, Y., Cheng, P., Chen, J. & Moscibroda, T., Mobility Modeling and Prediction in Bike-Sharing Systems. MobiSys, Singapore, 2016.

[8] Chen, L., Zhang, D., Wang, L., Yang, D., Ma, X., Li, S.,Wu, Z., Pan, G., Nguyen, T. & Jakubowicz, J., Dynamic Cluster-Based Over-Demand Prediction in Bike Sharing Systems. UbiComp, Heidelberg, Germany, 2016.

[9] Zhang, J., Pan, X., Li, M. &Yu, P., Bicycle-Sharing System Analysis and Trip Prediction. 17th IEEE International Conference on Mobile Data Management, Porto, Portugal, 2016.

[10] Citi Bike data, available at (accessed August 2018).

[11] Weather Underground, available at (accessed August 2018).

[12] Concert archives, available at (accessed August 2018).

[13] Basketball Reference, available at (accessed August 2018).

[14] New York Rangers, available at (accessed August 2018).

[15] Numbeo, available at (accessed August 2018).

[16] Wikipedia, available at (accessed August 2018).