State-of-the-art on clustering data streams

Abstract

Clustering is a key data mining task. This is the problem of partitioning a set of observations into clusters such that the intra-cluster observations are similar and the inter-cluster observations are dissimilar. The traditional setup where a static dataset is available in its entirety for random access is not applicable as we do not have the entire dataset at the launch of the learning, the data continue to arrive at a rapid rate, we can not access the data randomly, and we can make only one or at most a small number of passes on the data in order to generate the clustering results. These types of data are referred to as data streams. The data stream clustering problem requires a process capable of partitioning observations continuously while taking into account restrictions of memory and time. In the literature of data stream clustering methods, a large number of algorithms use a two-phase scheme which consists of an online component that processes data stream points and produces summary statistics, and an offline component that uses the summary data to generate the clusters. An alternative class is capable of generating the final clusters without the need of an offline phase. This paper presents a comprehensive survey of the data stream clustering methods and an overview of the most well-known streaming platforms which implement clustering.

Data stream clustering: a review

“…Four most commonly used data structures are feature vectors, prototype arrays, coreset trees and grids. Feature vectors keep the summary of the data instances, prototype arrays keep only a number of representative instances that exemplify the data, coreset trees keep the summary in a tree structure and grids keep the data density in the feature space (Ghesmoune et al, 2016;Mansalis et al, 2018;Silva et al, 2013).…”

“…Partitioning based algorithms have an easy implementation in general. StreamLSearch (O'Callaghan et al, 2002), incremental k-means (Ordonez, 2003), CluStream (Aggarwal et al, 2003), HP-Stream (Aggarwal et al, 2004), SWClustering (Zhou et al, 2008), StreamKM++ (Ackermann et al, 2012), strAP (Zhang et al, 2014) and CLARA (Kaufman and Rousseeuw, 1990) are partitioning based algorithms (Ghesmoune et al, 2016;Kumar, 2016;Mousavi et al, 2015). -Grid based algorithms use grid data structure.…”

Gene-Based Clustering Algorithms: Comparison Between Denclue, Fuzzy-C, and BIRCH

“…It is an incremental and dynamic clustering algorithm that follows a hierarchical clustering technique for databases by incrementally constructing a clustering feature (CF) tree, which is a subcluster of data points or better described as a tree-like representation of data points in a data set. 22 Best clustering is achieved by multi-scanning, and having more available memory which maximizes good result. 11 BIRCH is an incremental clustering algorithm that has 4 phases.…”

Optimizing Data Stream Representation: An Extensive Survey on Stream Clustering Algorithms

“…In Ghesmoune et al (2016) the authors discuss 19 algorithms and are among the first to highlight the research area of Neural Gas (NG) for stream clustering. However, only a single grid-based algorithm is discussed and other popular algorithms are missing.…”