Data stream mining has become a research area of some interest in recent years. The key challenge in data stream mining is extracting valuable knowledge in real time from a massive, continuous, dynamic data stream in only a single scan. Clustering is an efficient tool to overcome this problem. Data stream clustering can be applied in various fields such as financial transactions, telephone records, sensor network monitoring, telecommunications, website analysis, weather monitoring, and

more »

... s. Data stream clustering presents some challenges; it needs to be done in a short time frame with limited memory using a single-scan process. Moreover, because data stream outliers are hidden, clustering algorithms must be able to detect outliers and noise. In addition, the algorithms have to handle concept drift and detect arbitrary shaped clusters. Several algorithms have been proposed to overcome these challenges. This paper presents a review of five types of data stream clustering approaches: partitioning, hierarchical, density-based, grid-based and model-based. The different data stream clustering algorithms in the literature by considering their respective advantages and disadvantages are discussed.