变精度粗糙集挖掘技术的应用研究

Research of Application of Variable Precision Rough Set

粗糙集理论是一种处理模糊和不确定知识的一种新型数学工具,在很多领域取得了成功的应用。但是经典粗糙集理论处理的分类必须是完全正确的,在实际应用中,缺乏对噪声数据的适应能力,为了克服这个缺点,提出一种变精度的粗糙集模型,以适应实际应用的需要。对变精度粗糙集理论的数据预处理、属性约简、值约简和规则提取等问题进行了分析和研究,提出属性约简算法和基于求核值属性的归纳值约简算法,并将其运用于医疗系统的手术诊断数据表的数据挖掘分析过程中,所得到的实验结果与专家诊断结果基本吻合,取得了较好的实际应用效果。

Rough sets theory is a new mathematical tool to deal with problems on vagueness and uncertainty. It obtained many achievements in many fields in recent years. Unfortunately, it requires accurate classification in practice, it is lack of the capability of noise data processing. To deal with inconsistency in decision tables, the variable precision rough set model（VPRS）was developed by Prof. W. Ziarko in 1990s. This article introduces data mining process including data pre- process, attribute reduction, value reduction and rule generation based on the VPRS theory. These steps are used to deal with the data mining of medical diagnosis system. The theoretical result is basically identical with that of the qualitative experiment. Finally, the experimental results are used to show the application of VPRS theory and proves the rationality of the theory.