1 / 20

Real-Time Clustering of Large Geo-Referenced Data for Visualizing on Map

This project presents a real-time clustering algorithm for visualizing large amounts of geo-referenced data on a map. The proposed grid-growing algorithm overcomes the limitations of traditional clustering approaches and allows for efficient and accurate clustering of data points. The system is suitable for real-time applications with low bandwidth and offers a novel approach to web mapping systems.

hjeffrey
Télécharger la présentation

Real-Time Clustering of Large Geo-Referenced Data for Visualizing on Map

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Real-Time Clustering of Large Geo-Referenced Data for Visualizing on Map Presenter: Rushikesh Sane Course: LAMI, Spring 2019 6th February2019

  2. Real-Time Geo-Referenced Data Clustering

  3. VisualizingLARGEamount of data PredefinedFilters Clutter Slowness

  4. Clustering Problem • Number of Clusters • Minimizing SSE

  5. Whynotthese? DBSCAN K Means Hasdifferent objective Need to determine k in advance Hightimecomplexity O(N3) Hierarchical

  6. Grid-based Clustering

  7. Grid-basedSteps Grid Construction Initial Clustering Grid Construction

  8. Grid-basedapproach in clustering OverlappingClusters Cluster aftermerge

  9. Proposedapproach Client Server PrepareQuery and Parameters Queryfrom database • Grid construction • InitialClusters in thecell OverlapAcross Neighboringcells Displaycluster on Map Information of eachcell

  10. Experiments Processing time of 3 clustering approaches Clustering quality with 100 data points Veryfast! Comparison with Google cluster marker API

  11. A possiblegrid-growingclusteringalgorithm

  12. Steps • Grid Construction • Borderspassthroughmax & min co-ordinates • Initialseeds (randomor top m dense) • 2. Grid Growing • m initialseeds • 4 or 8 neighbors 3. Partitioning 4-neighborhood 8-neighborhood

  13. Experiments Litekmeans: Fastestimplementation of k means in matlab GEM: Greedyexpectation-maximizationalgorithm (gaussianmixturemodel) DBSCAN: Densitybasedclusteringalgorithm LSC: Landmarkspectralclustering (graphbasedalgorithm) PRS: Pairwiserandomswap GG: Grid Growing Leasttime Optimalstrategysuggestedbyauthors Time required Leastclustersdetected Maximum clustersdetected Efficiency of algorithm & number of clustersdetected

  14. Conclusion • Novel idea of web mapping system based on clustering to make real time queries. • First ever system to allow fetching query results up to 1M objects. • Suitable for real time applications having low bandwidth. • Theproposedgrid-growingalgorithmpossesestheadvantages of both k-means and DBSCAN. • Time complexityis O(nlogn). • Overcomesalldifficultiesrelated to geo-taggedrealtimeapplications.

  15. Thankyou Real-Time Clustering of Large Geo-Referenced Data for Visualizing on Map Mohammad Rezaei, PasiFranti Link: http://cs.uef.fi/sipu/pub/aece_2018_4_8.pdf A grid-growing clustering algorithm for geo-spatial data QinpeiZhaoa, Yang Shi Qin Liua, PasiFränti Link: http://cs.uef.fi/sipu/pub/Grid_growing_Zhao_2015.pdf OtherReferences: Gifs on slide 5 – https://giphy.com/gifs/dendrogram-dashee87githubio-scikit-pSNCWCEAsgrAs https://giphy.com/gifs/dbscan-dashee87githubio-scikit-OVJBPIB6oL3a0 https://giphy.com/gifs/needs-dashee87githubio-scikit-3NKUcoyBzkXQc

  16. Techniques to Visualize Data Technique for Data Reduction ShowingDensity Representation of Cluster ShowingDistribution Opening a Cluster Details on Demand

  17. Resistance to Problemsdue to Panning PanningProblem Solution: Initializethegridonlyone at thebeginning. A pointassigned to one box willremainassigned to it evenafterpanning. Butwetakeonlythoseobjects into considerationwhicharecompletelyorpartially Visible in thecurrentview.

  18. VoronoiDiagram In mathematics, a Voronoi diagram is a partitioning of a plane into regions based on distance to points in a specific subset of the plane. That set of points (called seeds, sites, or generators) is specified beforehand, and for each seed there is a corresponding region consisting of all points closer to that seed than to any other. These regions are called Voronoi cells.

  19. Whatif data point and grid-lineoverlap? Point could be placed in any of the adjacent cells. It’s a trivial issue. One can place it in more dense cell.

  20. SSE & number of clusters formula

More Related