摘要： 针对数据流的数据处理,给出了一种简单、实用的Agglomerative类直方图增量维护算法,该算法充分利用等宽直方图和变宽直方图优点,克服各自估算不够准确和维护复杂的缺点,并且能够以数据块为单位对直方图进行增量维护,通过合并与分裂直方图的桶,来平衡算法的计算速度、精确度和占用内存空间三者之间的关系。实验表明该算法对数据流数据的处理具有较强的实用性和有效性。

关键词: 数据流, 直方图, 近似算法, 批处理

Abstract: This paper presents a new algorithm for processing continuous data streams, which can incrementally maintain histograms with datablock. The histogram is an agglomerative histogram algorithm, which takes advantage from both equal-width histogram and variable–widthhistogram. By merging and splitting the buckets of histogram, tradeoffs among efficiency, accuracy and storage of the algorithm can be made. Theexperiment shows that the algorithm is effective and efficient for continuous data streams processing.

中图分类号: 

• TP311.13