Research Notes 3: Motivation of Data Stream Clustering Algorithms

Clustering algorithms separate a set of data points into two or more groups of similar points. Their performance indicators are the objective functions related to similarity.

Traditional clustering algorithms, such as K-means, hierarchical clustering, DBSCAN and SOM, revisit the data points several times to optimize the performance indicators.

However, the revisit of data points is not allowed in data stream clustering as the data volume grows along the time. This stimulates the development of data stream clustering algorithms. The short overview on data stream mining algorithms will be the subsequent post.