ligrong2001
在整理数据的时候,发现一些奇异值,听说可以用SAS做出茎叶图,就能看出哪些是奇异值。是不是可以自己私下删掉重新进行统计,不需要在论文里说明呀?我刚刚进入数理统计的世界,还不太清楚是怎么回事。请各位大侠指教。
rtist
搜一下论坛里的帖子吧,这种问题我都快回答腻了。
绝大多数情况下都不能删除。
Statsfu
you should try both cases: with outliers or without outliers. Then compare the result. SAS can help you clean the data.
It is important to study the outliers carefully before you neglect them, especially for the researchers. They might be caused by some unknown factors that the scientists are most interested in.
park
还是先做一下散点图,看有没有异常点和线性趋势,做了相关分析后,发现异常值,比较去掉异常值与未去掉异常值的结论的不同
TTG
[quote]引用第2楼Statsfu于2007-08-23 23:52发表的“”:
you should try both cases: with outliers or without outliers. Then compare the result. SAS can help you clean the data.
It is important to study the outliers carefully before you neglect them, especially for the researchers. They might be caused by some unknown factors that the scientists are most interested in.[/quote]
赞同,异常值有时藏着新发现呢
ligrong2001
[quote]引用第1楼rtist于2007-08-23 22:55发表的“”:
搜一下论坛里的帖子吧,这种问题我都快回答腻了。
绝大多数情况下都不能删除。[/quote]
我搜索过哦,可是没有搜出来。不知是不是方法不对,还是没有把问题说清楚。呵呵......
刚刚入这个论坛,花了一点时间看论坛的设置,可是还是耽误别人时间了。不好意思。
hexm26
[quote]引用第2楼Statsfu于2007-08-23 23:52发表的“”:
you should try both cases: with outliers or without outliers. Then compare the result. SAS can help you clean the data.
It is important to study the outliers carefully before you neglect them, especially for the researchers. They might be caused by some unknown factors that the scientists are most interested in.[/quote]
这个建议很好,对比一下包含和不包含outliers的结果很重要。不过,我真的很怀疑SAS可以帮助clean the data。