一种基于可变滑动窗口的数据流分段聚类算法

数据流的应用越来越广泛,数据流挖掘成为数据挖掘的重点研究方向之一。在分析各种数据流聚类算法的基础上,提出了一种基于可变滑动窗口的数据流分段聚类算法。算法以时间序列数据流模式表示技术为参考,以去除噪音和压缩数据为目的,实现了数据流的特征提取和概要存储。实验表明,算法具有低时空复杂度、自适应等特点。

A Piecewise Aggregate Approximation Algorithm of Data Stream based on Variable Sliding Window

The application of Data stream is more widely, the data stream mining became one focus of data mining research directions. Based on the analysis of various data stream clustering algorithm,this paper proposes a piecewise aggregate approximation algorithm of data stream based on variable sliding window. The algorithm through time series data stream pattern representation technology as the reference, with the purpose of removing noise and compressing data, it achieves the data stream feature extraction and summary storage. Experiments show that the algorithm has a low time and space complexity, adaptive characteristics.