Spatiotemporal Clustering Analysis of Bicycle Sharing System with Data Mining Approach

The main objective of this study is to explore the spatiotemporal activities pattern of bicycle sharing system by combining together temporal and spatial attributes variables through clustering analysis method. Specifically, three clustering algorithms, i.e., hierarchical clustering, K-means cluster...

Full description

Saved in:
Bibliographic Details
Published inInformation (Basel) Vol. 10; no. 5; p. 163
Main Authors Ma, Xinwei, Cao, Ruiming, Jin, Yuchuan
Format Journal Article
LanguageEnglish
Published Basel MDPI AG 01.05.2019
Subjects
Algorithms
bicycle sharing system
Bicycles
Bicycling
Brand loyalty
Cluster analysis
Clustering
Data mining
Docking
Land use
Literature reviews
Mobile commerce
point of interest (POI)
smart card data
Smart cards
Spatial data
spatiotemporal activities pattern
Stations
Statistical analysis
Vector quantization
Australia
Beijing China
China
Online AccessGet full text

Cover

More Information
Summary:The main objective of this study is to explore the spatiotemporal activities pattern of bicycle sharing system by combining together temporal and spatial attributes variables through clustering analysis method. Specifically, three clustering algorithms, i.e., hierarchical clustering, K-means clustering, expectation maximization clustering, are chosen to group the bicycle sharing stations. The temporal attributes variables are obtained through the statistical analysis of bicycle sharing smart card data, and the spatial attributes variables are quantified by point of interest (POI) data around bicycle sharing docking stations, which reflects the influence of land use on bicycle sharing system. According to the performance of the three clustering algorithms and six cluster validation measures, K-means clustering has been proven as the better clustering algorithm for the case of Ningbo, China. Then, the 477 bicycle sharing docking stations were clustered into seven clusters. The results show that the stations of each cluster have their own unique spatiotemporal activities pattern influenced by people’s travel habits and land use characteristics around the stations. This analysis will help bicycle sharing operators better understand the system usage and learn how to improve the service quality of the existing system.
ISSN:2078-2489
2078-2489
DOI:10.3390/info10050163