基于模式聚类的数据预处理方法应用研究

Study of the Data Pretreatment Method Based on Pattern Clustering

在分析数据预处理的意义基础上,提出了一种基于最大距离算法的模式聚类的数据预处理方法。该方法不依赖于任何数学模型,通过对某造纸厂大量数据的仿真处理,研究表明本文提出的方法能在保留原始数据的有用信息的基础上剔除冗余数据,侦破过失误差,减少随机误差。

Based on the analysis of the meaning of data pretreatment, this paper gives method of a data pretreatment based on the max distance arithmetic which is one of methods of pattern clustering. This method does not depend on math model. We have collected a lot of spot data of one paper mill and do much simulation. The result shows this method can eliminate gross errors and keep the valid information and minimize stochastic errors.