不同属性加权方法对冠心病数据分类效果的研究  

Study of different attribute weighting methods on coronary heart disease data for classification

作　　者：卢刚[1] 邓霖[1] 郑刚[1] 

机构地区：[1]天津理工大学计算机科学与技术学院,天津300191

出　　处：《天津理工大学学报》2008年第3期14-17,共4页Journal of Tianjin University of Technology

基　　金：天津市高等学校科技发展基金(20061009)

摘　　要：本文主要研究的是系统重构方法和层次分析方法在属性加权过程中的作用.根据背传神经网络获得分类正确率被作为比较标准.文中使用的数据是冠心病病例数据,直接来源于医院临床环节.实验表明,系统重构算法的加权数据的分类正确率要高于层次分析算法的加权结果,且两种加权方法的分类正确率都高于为加权的原始数据.The paper studies the attribute weighting effect of system reconstruction （SR） and Analytic hierarchy Process （AHP） method. The classification correction rate of diagnosis by one of artificial neural network algorithm, back propagation network （BP） is selected as comparison standard. The data used in the study is a coronary heart disease database, they are directly collected from clinical hospital for study. Experiment shows that the classification correction rate of weighting data by system reconstruction by is higher than that of weighting by analytic hierarchy process, and the classification correction rates of both weighted data are higher than that of original data.

关 键 词：属性 属性加权 系统重构算法 层次分析法 

分 类 号：TP311.1[自动化与计算机技术—计算机软件与理论]