A new trajectory clustering algorithm based on TRACLUS

Most trajectory clustering algorithms including the famous TRACLUS require the setting of two input parameters and are sensitive to input parameters. Incorrect setting may cause the algorithm to produce the wrong clusters. Aiming at this vulnerability, we propose a Shielding Parameters Sensitivity T...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:2012 2nd International Conference on Computer Science and Network Technology S. 783 - 787
1. Verfasser: Jiashun, Chen
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 01.12.2012
Schlagworte:
Clustering algorithms
Data mining
Depression
High dimensional data
input parameters
Merging
Partitioning algorithms
reachable plot
Sensitivity
Sorting
Trajectory
trajectory clustering
Uncertainty
ISBN:1467329630, 9781467329637
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Most trajectory clustering algorithms including the famous TRACLUS require the setting of two input parameters and are sensitive to input parameters. Incorrect setting may cause the algorithm to produce the wrong clusters. Aiming at this vulnerability, we propose a Shielding Parameters Sensitivity Trajectory Clustering algorithm named SPSTC. Firstly, we present some definitions about the core distance and reachable distance of line segment, according to which generates cluster sorting. Secondly, reachable plots of line segment sets are constructed according to cluster sorting and reachable distance. Thirdly, parameterized sequence is extracted according to reachable plot, and then the final trajectory clustering based on parameterized sequence is acquired. Parameterized sequence represents inner clustering structure of trajectory data. The experimental results on real and synthetic trajectory data demonstrate that the SPSTC algorithm reduces the sensitivity to input parameters remarkably and improves the efficiency of trajectory clustering while keeps the quality of trajectory clustering.
ISBN:1467329630
9781467329637
DOI:10.1109/ICCSNT.2012.6526048