A Multiscale Approach for Free-Float Bike-Sharing Electronic Fence Location Planning: A Case Study of Shenzhen City

Table 4

Evaluation of clustering methods.

Clustering methods	Number of clusters	Silhouette coefficient	Noise ratio (%)	Running times (s)
The graded clustering	86	−0.199433333	17.92713179	1.25679
DBSCAN	92	−0.3724	25.11887779	2.3274
K-means clustering	30	0.6418	—	4.0305

The bold values indicate the superiority of the proposed clustering method “the graded clustering” under the three indicators.