CONTACT US: Contact info
|
Citation |
Kim K. J. Transp. Geogr. 2018; 66: 309-320. |
|
Copyright |
(Copyright © 2018, Elsevier Publishing) |
|
DOI |
|
PMID |
unavailable |
|
Abstract |
Public bicycle systems are widely spread across many cities worldwide. 'Tashu', a public bicycle sharing system in Daejeon, was installed in 2009 and it is one of the well-established public bike-sharing systems in South Korea. Previous studies in the literature found that in general, bicycling is affected by weather conditions and temporal characteristics. However, the degrees of impacts or the signs of effects may be different depending on the stations. Therefore, this study investigated the different effects of weather conditions and temporal characteristics according to the characteristics of the stations at the station level analysis in addition to the system level analysis. For the cost-effective station level analysis, clustering analysis was utilized to find out the groups of the stations with the similar properties. Moreover, temperature humidity index (THI) and the indicator variable of heatwaves were introduced to consider the interaction between temperature and humidity and measure the influence of high temperature, which has been rarely considered. In the system level analysis, the results showed that the selected factors have the different influence over the different time periods within a day. Especially, scorching heat and non-working days differently affect the demand for public bikes by hours. Also, it was observed that high temperature over 30°C reduces the bicycle usage, which revealed the necessity of taking into account not only severe colds but also heatwaves in the prediction of the demand. By clustering analysis, the stations were partitioned into the three clusters. One cluster shows the strong peak in the morning while two others have peaks in the evening. The effects of weather conditions and non-working days on the demand for public bicycles were different depending on the clusters, which seemed to be related to the main purposes of bike usage in the clusters. Language: en |
|
Keywords |
Clustering analysis; Negative binomial regression analysis; Public bicycle system |