Towards Mobilitybased Clustering Siyuan Liu Yunhuai Liu Lionel
Towards Mobility-based Clustering Siyuan Liu, Yunhuai Liu, Lionel M. Ni, Jianping Fan, Minglu Li Proceedings of the 16 th ACM SIGKDD international conference on Knowledge discovery and data mining(2010) 2011년 11월 21일 김호영
Outline Introduction Related work Mobility-based clustering Conclusion Page 2
Introduction Smart City Application – City monitoring and management – Identify hot spot of moving vehicles in urban area Data collection, storage and mining – Vehicle GPS data set – Mobile phone networks data set Hot Spot detection in the city – Traffic jam – Exhibitions – Commercial promotions Page 3
Introduction Crowded spots in the city Vehicle instant locations of sample taxis at 13: 00 PM on 12 th Dec, 2006 Hot Spot : Area of high crowdedness of vehicles Page 4
Introduction Data Set – Ideal • Information of all vehicles in the city – Real • Only a sample set of all vehicles – Taxi GPS Data(ID, Location, Speed, Time, Direction, Status) – Approximately 0. 3% of the 2, 000 vehicles in Shanghai(5631 vehicles) – How can use limited sample data set to detect hot spot in the city? Page 5
Introduction Challenges Dense Data Set High Crowdedness Vehicle crowdedness distribution and instant locations of sample taxis at 14: 00 PM on 12 th Dec, 2006 Could density-based clustering handle it? Page 6
Introduction Methodology – Observation • The low speed may indicate that the area is crowded – Method • Mobility-based clustering • Study the speed(mobility) instead of the density – Employ moving object as sensors Page 7
Related Work Static Objects • • • Moving Objects • Raw data-based methods • Feature-based methods • Model-based methods Problem • What if the mobility is high? • What if the density is poor? • What if the location is lossy? Partitioning methods Hierarchical methods Density-based methods Grid-based methods Model-based methods Page 8
Mobility-based clustering Object mobility model – Road network grid – Interpolation Spot crowdedness model – Linear crowdedness function – Statistical crowdedness function Page 9
Mobility-based clustering Spot crowdedness in practice Characterizing spots Hot spots and hot regions Spot crowdedness Profiling sensor object Temporal hot spots Evolutionary hot regions Page 10
Mobility-based clustering Characterizing spots Page 11
Mobility-based clustering Sensor object profiling Page 12
Mobility-based clustering Hot spots and hot regions – Hot spot – Hot region Not hot spot, even dense sample points Hot spot, even sparse sample points Page 13
Mobility-based clustering Temporal hot spots Page 14
Mobility-based clustering Evolutionary hot regions – Area difference ratio – Corwdedness difference ratio Page 15
Mobility-based clustering Crowdedness model validation – Taxis traces – Buses traces Page 16
Mobility-based clustering validation Page 17
Conclusion and future work Contirbutions – Mobility-based clustering model – Key factors on spot crowdedness – Hot spots and hot regions Future work – More accurate speed information – More accurate location information Page 18
- Slides: 18