Identification of urban regions' functions in Chengdu, China, based on vehicle trajectory data

Data about human trajectories has been widely used to study urban regions that are attractive to researchers and are considered to be hotspots. It is difficult, however, to quantify the function of urban regions based on the varieties of human behavior. In this research, we developed a clustering me...

Full description

Saved in:
Bibliographic Details
Published in:PloS one 2019-04, Vol.14 (4), p.e0215656-e0215656
Main Authors: Gao, Qingke, Fu, Jianhong, Yu, Yang, Tang, Xuehua
Format: Article
Language:English
Subjects:
Analysis
Cellular telephones
China
Cities
Cluster Analysis
Clustering
Correlation coefficient
Correlation coefficients
Data points
Editors
Engineering
Geography
Human behavior
Humans
Inflow
Information science
International conferences
Knowledge discovery
Methods
Models, Statistical
Motor Vehicles - statistics & numerical data
Normal Distribution
Probabilistic models
R&D
Remote sensing
Remote Sensing Technology
Research & development
Statistical methods
Trajectories
Travel
Travel - statistics & numerical data
Urban areas
Urban Population - statistics & numerical data
Urban studies
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Data about human trajectories has been widely used to study urban regions that are attractive to researchers and are considered to be hotspots. It is difficult, however, to quantify the function of urban regions based on the varieties of human behavior. In this research, we developed a clustering method to help discover the specific functions that exist within urban regions. This method applies the Gaussian Mixture Model (GMM) to classify regions' inflow and trip count characteristics. It regroups these urban regions using the Pearson Correlation Coefficient (PCC) clustering method based on those typical characteristics. Using a large amount of vehicle trajectory data (approximately 1,500,000 data points) in the Chinese city of Chengdu, we demonstrate that the method can discriminate between urban functional regions, by comparing the proportion of surface objects within each region. This research shows that vehicle trajectory data in different functional urban regions possesses different time-series curves, while similar types of functional regions can be identified by these curves. Compared with remote sensing images and other statistical methods which can provide only static results, our research can provide a timely and effective approach to determine an urban region's functions.
ISSN:1932-6203
1932-6203
DOI:10.1371/journal.pone.0215656