数据处理中异常值的取舍

• 作 者: 马林茂
• 机 构: 中国疾病预防控制中心公共卫生检测与信息服务中心,
• 出 处: 疾病监测
• 发表日期: 2010年12期
• 关键词: 流行病学 数据处理 异常值

在流行病学调查和实验室资料的处理中有时会出现少数过大或过小的极端值,这种数据不但会影响结果而且使人怀疑发生了错误。对可疑值的处理首先应该反复检查、核对、寻找出错的原因,以便加以纠正。若找不出错误的任何原因,也不能将这种数值随意删除,一般处理的方法是增加观察次

Occasionally, there may be a few extremes that are either too large or too small for epidemiological investigations and laboratory data processing, and this data not only affects the outcome but also gives rise to suspicions of mistakes. The treatment of suspicious values ​​should first be repeated inspection, check, find the cause of the error, in order to be corrected. If you can not find any reason for the error, you can not delete this value at random, the general approach is to increase the number of observations