Please enable JavaScript.
Coggle requires JavaScript to display documents.
Cluster B: Conventional clustering algorithms (6. Graph Theory (Node…
Cluster B: Conventional clustering algorithms
6. Graph Theory
Node represents data and edge represents relationship
High clustering efficiency, accuracy
CLICK, MST
Graph complexity increases time complexity
8. Fractal Theory
Changed inner cluster data doesn't influence fractal dimension quality
FC
High clustering efficiency and scalability, effectively deals with outliers, arbitrary data and high dimension
Clustering sensitive to parameters
9. Model
Particular model for each cluster
Statistical Methods - COBWEB, GMM and CLASSIT, Neural Network Methods - SOM , ART and EM
Models extend significant advantages to specific areas
High time complexity, clustering sensitive to model parameters
7. Grid Theory
Original data changed to definite size grid for clustering
STING, CLIQUE
Mesh size sensitive clustering, high calculation efficiency at reduced clustering quality and accuracy
Low time complexity, high scalability, parallel processing and incremental update suitable