Data Stream Clustering Algorithm Research Articles

Research on Dynamic Data Streams Clustering Algorithm –Pdstream Based on PCA and Density

The research on data streams clustering has become a focus in the field of data streams mining. Because the number of data streams is too large, and CPU of the computer has limited memory and time, it’s difficult to carry out clustering quickly and effectively. For that problem, we design an improved clustering algorithm for dynamic data streams based on principal component analysis and density. The PDStream algorithm effectively overcomes the shortcomings of the STREAM algorithm controlled by historical data and the CluStream algorithm is difficult to describe non-spherical and out "old data", resulting in huge amount of data. In the course of the experiment, we compare with the STREAM algorithm, the PDStream algorithm shows the superiority of handling mass data and the characteristics of high-quality clustering.

Density grid-based data stream clustering algorithm over sliding window

This paper introduced a density grid-based data stream clustering algorithm.Through the introduction of the "subject degree",the traditional density grid-based clustering algorithm for data stream was improved by taking the data points within the grid as the grid density,thereby resolving the problem of data points belonging to two classes in one grid as well as the treatment of boundary points.Therefore,not only the high efficiency of the grid-based algorithm was utilized,but also the clustering accuracy was raised significantly.

HE-Tree: a framework for detecting changes in clustering structure for categorical data streams

Analyzing clustering structures in data streams can provide critical information for real-time decision making. Most research in this area has focused on clustering algorithms for numerical data streams, and very few have proposed to monitor the change of clustering structure. Most surprisingly, to our knowledge, no work has been proposed on monitoring clustering structure for categorical data streams. In this paper, we present a framework for detecting the change of primary clustering structure in categorical data streams, which is indicated by the change of the best number of clusters (Best K) in the data stream. The framework uses a Hierarchical Entropy Tree structure (HE-Tree) to capture the entropy characteristics of clusters in a data stream, and detects the change of Best K by combining our previously developed BKPlot method. The HE-Tree can efficiently summarize the entropy property of a categorical data stream and allow us to draw precise clustering information from the data stream for generating high-quality BKPlots. We also develop the time-decaying HE-Tree structure to make the monitoring more sensitive to recent changes of clustering structure. The experimental result shows that with the combination of the HE-Tree and the BKPlot method we are able to promptly and precisely detect the change of clustering structure in categorical data streams.

A Holistic Approach for Resource-aware Adaptive Data Stream Mining

Mining data streams is a field of increasing interest due to the importance of its applications and dissemination of data stream sources. Most of the streaming techniques developed so far have not addressed the need for resource-aware computing in data stream analysis. The fact that streaming information is often generated or received onboard resource-constrained computational devices such as sensor nodes and mobile devices motivates the need for resource-awareness in data stream processing systems. In this paper, we propose a generic framework that enables resource-awareness in streaming computation using algorithm granularity settings in order to change the resource consumption patterns periodically. This generic framework is applied to a novel threshold-based micro-clustering algorithm to test its validity and feasibility. We have termed this algorithm as RA-Cluster. RA-Custer is the first data stream clustering algorithm that can adapt to the changing availability of different resources. The experimental results show the applicability of the framework and the algorithm in terms of resource-awareness and accuracy.

Data stream clustering by fast density-peak-search

Data stream mining has recently been studied extensively in the literature. Many clustering algorithms were proposed to handle massive streams of data. However, many of these algorithms may not be as efficient as one desires for data streams, as they typically require a number of iterations in their implementations. In this paper, we will propose a new data stream clustering algorithm, based on the fast density-peak-search method. It does not require any iterations in its implementation, and therefore is most suitable for large streams of data. The comparisons of numerical illustration as well as a real example will be made with other alternative data stream algorithms.