子集不平衡(异型复制复制)来完成或平衡r中的数据集

Question

我有一个数据集,重复次数不等.我希望通过删除那些不完整的条目(即复制小于最大值)来对数据进行子集化.只是一个小例子:

set.seed(123)
mydt <- data.frame (name= rep ( c("A", "B", "C", "D", "E"), c(1,2,4,4, 3)), 
                   var1 = rnorm (14, 3,1), var2 = rnorm (14, 4,1))
 mydt
       name     var1     var2
1     A 2.439524 3.444159
2     B 2.769823 5.786913
3     B 4.558708 4.497850
4     C 3.070508 2.033383
5     C 3.129288 4.701356
6     C 4.715065 3.527209
7     C 3.460916 2.932176
8     D 1.734939 3.782025
9     D 2.313147 2.973996
10    D 2.554338 3.271109
11    D 4.224082 3.374961
12    E 3.359814 2.313307
13    E 3.400771 4.837787
14    E 3.110683 4.153373

摘要(mydt)

name       var1            var2      
 A:1   Min.   :1.735   Min.   :2.033  
 B:2   1st Qu.:2.608   1st Qu.:3.048  
 C:4   Median :3.120   Median :3.486  
 D:4   Mean   :3.203   Mean   :3.688  
 E:3   3rd Qu.:3.446   3rd Qu.:4.412  
       Max.   :4.715   Max.   :5.787 

我想从数据中删除A,B,E,因为它们不完整.因此预期输出:

name     var1     var2
4     C 3.070508 2.033383
5     C 3.129288 4.701356
6     C 4.715065 3.527209
7     C 3.460916 2.932176
8     D 1.734939 3.782025
9     D 2.313147 2.973996
10    D 2.554338 3.271109
11    D 4.224082 3.374961

请注意数据集很大,以下可能不是一个选项:

mydt[mydt$name == "C",]
mydt[mydt$name == "D", ]

Answer 1

这是一个解决方案data.table:

library(data.table)
DT <- data.table(mydt, key = "name")
DT[, N := .N, by = key(DT)][N == max(N)]
#    name     var1     var2 N
# 1:    C 3.070508 2.033383 4
# 2:    C 3.129288 4.701356 4
# 3:    C 4.715065 3.527209 4
# 4:    C 3.460916 2.932176 4
# 5:    D 1.734939 3.782025 4
# 6:    D 2.313147 2.973996 4
# 7:    D 2.554338 3.271109 4
# 8:    D 4.224082 3.374961 4

.N为您提供每组案例的数量,并使用复合查询data.table的选项,您可以根据您想要的新变量的条件立即进行子集化.

基数R也有几种方法,其中最明显的是table:

with(mydt, mydt[name %in% names(which(table(name) == max(table(name)))), ])

可能不太常见,但在data.table建议方法上类似,是使用ave():

counts <- with(mydt, as.numeric(ave(as.character(name), name, FUN = length)))
mydt[counts == max(counts), ]