前置粗糙分类器的基于关联规则的Logistic回归在脑梗塞发病因素分析中的应用  

The application of Logistic regression in cerebral infraction disease detection based on association rules with pre-rough classifier

作　　者：杨丽[1] 徐德生[1] 李长青[1] 田文生[2] 

机构地区：[1]内蒙古工业大学管理学院管理科学系,内蒙古呼和浩特010051 [2]内蒙古呼和浩特市第一医院,内蒙古呼和浩特010062

出　　处：《数字技术与应用》2012年第5期46-48,共3页Digital Technology & Application

摘　　要：研究西北少数民族地区引起脑梗塞发病的环境、习俗、健康状况和病史等复杂因素,从中找出关键因素进行预防。采用粗糙集进行属性约减,然后进行关联规则挖掘,并对挖掘结果进行Logistic回归。提出的模型解决了属性冗余和置信度框架常常产生过多的规则的缺陷。研究发现有其他脑血管病史、喝酒、季节变化是脑梗塞发病的显著性因素。对于西北少数民族地区居民,患过其他脑血管疾病、饮酒、季节变化都是脑梗塞发病的重要影响因素,而民族是非独立性显著因素。Environment,customs and health status in northwest minority areas have been studied.We found the critical factors to prevent cerebral infraction.First rough sets theory bad been used to reduce the attributes,secondly association roles had been used,finally logistic regression model had been used. The model solved the shortcomings of too many rules that caused by attribute redundancy and reliability framework.The results show that the history of other cerebrovascular disease,alcohol consumption and seasonal change are the significant factors of cerebral infraction.For the northwest minority residents,the history of other cerebrovascular disease,alcohol consumption and seasonal change are important factors of cerebral infraction,while nation is an independent non-significant factor.

关 键 词：关联规则 数据挖掘 LOGISTIC回归 脑梗塞 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]