A carregar...
Publicação
Station segmentation of Lisbon bicycle sharing system based on users demand and supply
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 534.85 KB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Bike-sharing systems are well known in the sustainable mobility field and have several aspects that
need optimization and improvement. One of the most relevant aspects is station segmentation based
on user demand and supply, and it is the focus of the thesis. The segmentation work has an enormous
potential to reduce complexity in predicting the bicycle demand and supply, thus improving the overall
quality of service.
Several machine learning algorithms were used to investigate the aforementioned segmentation
task. This work considers two popular and well-known clustering algorithms to extract and analyze
interesting patterns, like the difference between arrivals and departures throughout time and stations:
the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) and the hierarchical
clustering.
The algorithms are applied to the specific case of GIRA, the bicycle sharing system (BSS) of the city
of Lisbon. The obtained results suggest that considering the variables under analysis, the optimal
number of clusters to be used in a second phase of the BSS optimization (demand and supply forecast)
is the same as the number of stations in the Lisbon BSS. The results are very insightful and allow future
work to focus either on the demand forecast or the enrichment of the variables under study.
Descrição
Project Work presented as the partial requirement for obtaining a Master's degree in Information Management, specialization in Knowledge Management and Business Intelligence
Palavras-chave
Machine learning Timeseries segmentation Bike-sharing systems Sustainable mobility