一种基于可变滑动窗口的数据流分段聚类算法  

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

作　　者：栗磊[1] 周云霞[1,2] 张国强 

机构地区：[1]东北石油大学 [2]大庆普华科创信息工程有限公司,大庆163318

出　　处：《科学技术与工程》2014年第9期211-214,226,共5页Science Technology and Engineering

摘　　要：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。The application of data stream is more widely, the data stream mining became one focus of data min- ing research directions. Based on the analysis of various data stream clustering algorithm, a piecewise aggregate ap- proximation algorithm of data stream based on variable sliding window is proposed. The algorithm through time se- ries data stream pattern representation technology as the reference, with the purpose of removing noise and com- pressing data, it achieves the data stream feature extraction and summary storage. Experiments show that the algo- rithm has a low time and space complexity, adaptive characteristics.

关 键 词：数据流 数据流聚类 滑动窗口 

分 类 号：N311.11[自然科学总论]