r 通过 ddply 从具有两个标识符的数据框中删除异常值

Question

首先，我应该声明我对R语言不是很有经验。我有一个大型长格式数据框，以下面的df为例，有 3 列：Group、ID和dat。我想删除每个"组 ID"中的异常值(或者更确切地说是用平均值替换(。

Group = c("1","1","2","2","3","3","1","1","2","2","3","3","1","1","2","2","3","3","1","1","2","2","3","3")
ID = c("Eb","Eb","Eb","Eb","Eb","Eb","Sd","Sd","Sd","Sd","Sd","Sd","Re","Re","Re","Re","Re","Re","Tf","Tf","Tf","Tf","Tf","Tf")
dat = c(2,3,4,5,6,7,8,9,1010,11,12,13,1,2,3,-10000,5,6,4,3,2,7,6666,5)
df = data.frame(Group,ID,dat)

我的基本方法(不起作用(如下(我已经尝试了此代码的几次迭代(：

library(outliers)
library(plyr)
# Function to remove outliers
RmOurliFUN = function(x){
rm.outlier(x$dat, fill = TRUE)
}
# splitting data based on first Group, and then ID to apply the outlier removal
GroupSplit = function(x){ddply(x,"ID",RmOurliFUN)}
df2 = ddply(df1, "Group", GroupSplit)

我收到各种错误消息，但通常参数不是数字或逻辑的。我很确定我没有在嵌套>嵌套函数中正确调用dat列。 如何执行这样的操作？我愿意接受任何建议。

Answer 1

要删除每个唯一Group+ID组合中的异常值，您可以将函数直接添加到对ddply的调用中，然后重塑结果

library(outliers)
library(plyr)
library(reshape2)
#Make some new categories to have enough values for outlier detection
Group<-rep(c("a", "b"), each=12)
ID<-rep(c("c", "d"), each=6)
dat = c(2,3,4,5,6,7,8,9,1010,11,12,13,1,2,3,-10000,5,6,4,3,2,7,6666,5)
df1 = data.frame(Group,ID,dat)
df2<-ddply(df1, c("Group", "ID"), function(x) rm.outlier(x$dat, fill=TRUE))
#reshape and order the data
res<-melt(df2, id.vars=c("Group", "ID"), value.name = "dat")  
res<-arrange(res, Group, ID)[,-3]