Question

我需要从组内的每个观察中减去由组计算的平均值。其中具有挑战性的部分是我在数据框中有一个子分组，或两个级别：V5和V4。以下是我的data.frame结构的示例：

B = as.data.frame(matrix(
c(2,2,3,3,4,3,1,5,7,6,4,5,8,9,2,3,8,4,5,0,7,5,6,7,5,3,2,
"A","A","A","A","B","B","C","C","C",
"TRUE","TRUE","TRUE","TRUE","FALSE","FALSE","FALSE","FALSE","FALSE"),
nrow=9,ncol=5))

所以我的data.frame B看起来像这样：

  V1 V2 V3 V4    V5
1  2  6  5  A  TRUE
2  2  4  0  A  TRUE
3  3  5  7  A  TRUE
4  3  8  5  A  TRUE
5  4  9  6  B FALSE
6  3  2  7  B FALSE
7  1  3  5  C FALSE
8  5  8  3  C FALSE
9  7  4  2  C FALSE

因此，如果我按V5和V4平均得到一个新的data.frame，我称之为test，它考虑了多级分组：

test <- aggregate(. ~ B$V5+B$V4,data=B, mean)
> test
   B$V5 B$V4  V1       V2   V3 V4 V5
1  TRUE    A 2.5 4.500000 3.75  1  2
2 FALSE    B 3.5 4.000000 5.50  2  1
3 FALSE    C 4.0 3.666667 3.00  3  1

所以我正在努力解决的问题是从data.frame test中的原始观察中减去data.frame B中两级组的均值。直观地说，我假设有一个apply（）函数和某种条件语句可能，但它对我来说有点高级编码，我还在学习R.

Answer 1

以下是基础R的解决方案：

B <- data.frame(matrix(c(2,2,3,3,4,3,1,5,7,6,4,5,8,9,2,3,8,4,5,0,7,5,6,7,5,3,2), 9),
  V4=c("A","A","A","A","B","B","C","C","C"),
  V5=c("TRUE","TRUE","TRUE","TRUE","FALSE","FALSE","FALSE","FALSE","FALSE"))
B[1:3] <- lapply(B[1:3], function(x) x - ave(x, B$V4, B$V5, FUN=mean))
B

我使用了其他数据。在您的示例数据框中，所有列都是因子（您无法使用因子进行计算，例如mean(...)）。

Answer 2

我们可以使用data.table执行此操作。将'data.frame'转换为'data.table'（setDT(B)），按“V4”，“V5”分组，循环遍历data.table（.SD）的子集并获得差异对于每个组，每列的mean列为

library(data.table)
setDT(B)[, lapply(.SD, function(x) x- mean(x)), by = .(V4, V5)]

或者我们可以使用dplyr

library(dplyr)
B %>%
  group_by(V4, V5) %>%
  mutate_all(funs(.- mean(.)))
# A tibble: 9 x 5
# Groups: V4, V5 [3]
#          V1    V2         V3     V4     V5
#       <dbl> <dbl>      <dbl> <fctr> <fctr>
#1 -0.5000000  0.25  0.7500000      A   TRUE
#2 -0.5000000 -1.75 -4.2500000      A   TRUE
#3  0.5000000 -0.75  2.7500000      A   TRUE
#4  0.5000000  2.25  0.7500000      A   TRUE
#5  0.5000000  3.50 -0.5000000      B  FALSE
#6 -0.5000000 -3.50  0.5000000      B  FALSE
#7 -3.3333333 -2.00  1.6666667      C  FALSE
#8  0.6666667  3.00 -0.3333333      C  FALSE
#9  2.6666667 -1.00 -1.3333333      C  FALSE

假设前3列为numeric

数据

B <- structure(list(V1 = c(2, 2, 3, 3, 4, 3, 1, 5, 7), V2 = c(6, 4, 
5, 8, 9, 2, 3, 8, 4), V3 = c(5, 0, 7, 5, 6, 7, 5, 3, 2), V4 = structure(c(1L, 
1L, 1L, 1L, 2L, 2L, 3L, 3L, 3L), .Label = c("A", "B", "C"), class = "factor"), 
V5 = structure(c(2L, 2L, 2L, 2L, 1L, 1L, 1L, 1L, 1L), .Label = c("FALSE", 
"TRUE"), class = "factor")), .Names = c("V1", "V2", "V3", 
"V4", "V5"), row.names = c(NA, -9L), class = "data.frame")

从data.frame中的每个观察值中减去两级组均值

2 个答案:

数据