Skip to content

Hippoangel/DBS

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Dock-less Bike Sharing with Network

Dock-less Bike Sharing

abstract of Extreme unbalanced mobility network in bike sharing system(2020) by Zhiguo Zheng, Daqing Li

Bike sharing, such as Mobike, are thriving throughout the past three years. Mobike allows users to rent bikes left by previous users. This mobility mode generates self- organized bike distribution over the city. We analyze real data from Mobike bike sharing system in Beijing and Citibike bike sharing system in New York. In this paper, we study the spatial pattern of sharing activity from this new business model. We construct an Origin–Destination spatial network of bike sharing traffic flow based on trip data and study the characteristic of unbalance in the systems. The unbalance of nodes in Mobike bike mobility OD network is revealed to follows a power law distribution with different exponents on weekdays and weekends. We also found that there exists a scaling between maximum unbalance and network degree in both systems. These features of bike sharing can provide a better understanding of sharing mobility behaviors and help to develop more efficient management methods.

Data

Singapore (https://github.com/dkondor/bikesharing_data)

Shanghai (http://shanghai.sodachallenges.com/data.html?lang=en)

Beijing (https://biendata.com/competition/mobike/data/)

Los Angeles, San Francisco (https://datadryad.org/stash/dataset/doi:10.25338/B8X064)

Method

DBSCAN

Even Daqing Li group completed their researches successfully, there are big suspicious component that they build the network by grid.

I caught such a problems is from they formed network in grid way. The challenge is to find out natural dock- or station- of bike to solve every engaged problems otherwise grid-network ruined its property.

So, I tried to build real (nearly) network by adopting DBSCAN. And, also some inceptions about these problems

Power-law fitting

plus Daqing Li group insist that its network must be power-law -scale-free- neglecting the grid has only 100 or more, and it's difference between highest value and lowest was 100.

[Power-law distributions in empirical data](http plus they insist that its network must be power-law -scale-free- neglecting the grid has only 100 or more, and it's difference between highest value and lowest was 100. s://arxiv.org/abs/0706.1062)(Clauset et al., 2007)

Clustering Validation

Evaluation (or "validation") of clustering results is as difficult as the clustering itself. Popular approaches involve "internal" evaluation, where the clustering is summarized to a single quality score, "external" evaluation, where the clustering is compared to an existing "ground truth" classification, "manual" evaluation by a human expert, and "indirect" evaluation by evaluating the utility of the clustering in its intended application. (Wikipedia)

Based on Performance evaluation of some clustering algorithms and validity indices(U. Maulik et al., 2002), built a code and apply it to various hyper-parameters.

image

CH

XB

I-index

Dunn's indices

Scale

By Emergence of scaling in dockless bike-sharing systems(Ruiqi Li et al., 2022) , I could get a insight about scaling of human mobility in city.

image

And this paper also adopted the grid-based network. To reinforce my method's rationality I will check out the emergence of scaling through DBSCAN.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published