Summary of Clustering algorithm

Read some papers, a summary, share to everyone!

Clustering algorithm can be divided into the following categories: 1 The main idea of partitioning method is: Given the number of partitions to be constructed K, randomly select K objects in the database, each object represents the average or center of a class, and divides it into the nearest class based on the distance from the remaining object to the center of the class.    Then recalculate the center of each class, repeating the process until all objects are no longer allocated. Typical partitioning methods include: K-means, K-Medoids, CLARA, Clarans, FCM, etc. 2) The main idea of hierarchical approach based on hierarchical method is: to decompose the set of a given data object in a hierarchical way.    According to the formation of hierarchical decomposition, the hierarchical method is divided into condensed method (bottom-up) and splitting method (top-down). The algorithm based on hierarchical clustering mainly includes: Cure algorithm, BIRCH, rock algorithm, Chameleon and so on. 3) The main idea of density-based methods is to continue clustering as long as the density of the adjacent area exceeds a given threshold beforehand.    This way, you can find classes of arbitrary shapes and filter out "noise" points. Density-based clustering algorithms mainly include: DBSCAN, OPTICS, Denclue, and so on. 4) The main idea of grid-based method is to quantify the object space into a finite number of cells, and to form a grid structure to perform all cluster operations on the grid structure. Grid-based methods are: STING, Clque, Wavecluster. 5) The main idea of model-based approach is to assume a model for each class and find the optimal fit of the data to a given model.     Model-based algorithm localization clustering is achieved by constructing a density function that reflects the spatial distribution of data points, and it also automatically determines the number of clusters by standard statistics, and considers noise data or outliers to generate robust clustering methods. The algorithm based on the model method mainly includes: Cobweb, Classit and so on.

Summary of Clustering algorithm