我正在尝试按组获取某些列的加权行平均值。
library(tidyverse)
set.seed(1)
df <- data.frame(group = rep(LETTERS[3:4], each = 10),
x = runif(n = 10, min = 10, max = 15),
y = runif(n = 10, min = 100, max = 150),
z = runif(n = 10, min = 100, max = 150))
df
# group x y z
# 1 C 11.32754 110.2987 146.7353
# 2 C 11.86062 108.8278 110.6071
# 3 C 12.86427 134.3511 132.5837
# 4 C 14.54104 119.2052 106.2778
# 5 C 11.00841 138.4921 113.3610
# 6 C 14.49195 124.8850 119.3057
# 7 C 14.72338 135.8809 100.6695
# 8 C 13.30399 149.5953 119.1194
# 9 C 13.14557 119.0018 143.4845
# 10 C 10.30893 138.8723 117.0174
# 11 D 11.32754 110.2987 146.7353
# 12 D 11.86062 108.8278 110.6071
# 13 D 12.86427 134.3511 132.5837
# 14 D 14.54104 119.2052 106.2778
# 15 D 11.00841 138.4921 113.3610
# 16 D 14.49195 124.8850 119.3057
# 17 D 14.72338 135.8809 100.6695
# 18 D 13.30399 149.5953 119.1194
# 19 D 13.14557 119.0018 143.4845
# 20 D 10.30893 138.8723 117.0174
要获得x、y、z的粗行平均值，我可以这样做:
df %>%
mutate(rmean = pmap_dbl(list(x, y, z), ~mean(c(...))))
但我想用这些权重来衡量它们
dfweight <- data.frame(group = c("C", "C", "C",
"D", "D", "D"),
cat = c("x", "y", "z",
"x", "y", "z"),
weights = c(.2, .7, .1,
.4, .1, .5))
# group cat weights
# 1 C x 0.2
# 2 C y 0.7
# 3 C z 0.1
# 4 D x 0.4
# 5 D y 0.1
# 6 D z 0.5
我想我应该先提取权重:
dfweight_split <- lapply(split(dfweight, dfweight$group), function (x) x$weights)
dfweight_split
# $C
# [1] 0.2 0.7 0.1
# $D
# [1] 0.4 0.1 0.5
但是我不确定如何pmap/map这些？
df %>%
group_by(group) %>%
mutate(wmean = pmap_dbl(list(x, y, z), ~weight.mean(c(..., dfweight_split))))
#OR
df %>%
group_by(group) %>%
mutate(wmean = map2(list(x, y, z), dfweight_split, ~weight.mean(.x, .y)))
也很高兴看到基本解决方案。类似的帖子是here .
谢谢
最佳答案
如果我们想使用pmap ，确保“dfweight”数据列也位于同一数据集中。一个选项是使用 pivot_wider reshape 为宽幅。 ，然后进行连接 ( right_join ) 并使用 pmap要循环行，请使用符号 .. 以相同的顺序提取列元素在索引之前，将它们传递为 vector weighted.mean 中的参数在 mutate 中创建列
library(dplyr)
library(purrr)
library(tidyr)
library(stringr)
dfweight %>%
pivot_wider(names_from = cat, values_from = weights) %>%
rename_at(-1, ~ str_c(., '_weight')) %>%
right_join(df) %>%
mutate(wmean = pmap_dbl(select(., -group),
~ weighted.mean(c(..4, ..5, ..6), c(..1, ..2, ..3)))) %>%
select(-ends_with('weight'))
# A tibble: 20 x 5
# group x y z wmean
# <chr> <dbl> <dbl> <dbl> <dbl>
# 1 C 11.3 110. 147. 94.1
# 2 C 11.9 109. 111. 89.6
# 3 C 12.9 134. 133. 110.
# 4 C 14.5 119. 106. 97.0
# 5 C 11.0 138. 113. 110.
# 6 C 14.5 125. 119. 102.
# 7 C 14.7 136. 101. 108.
# 8 C 13.3 150. 119. 119.
# 9 C 13.1 119. 143. 100.
#10 C 10.3 139. 117. 111.
#11 D 11.3 110. 147. 88.9
#12 D 11.9 109. 111. 70.9
#13 D 12.9 134. 133. 84.9
#14 D 14.5 119. 106. 70.9
#15 D 11.0 138. 113. 74.9
#16 D 14.5 125. 119. 77.9
#17 D 14.7 136. 101. 69.8
#18 D 13.3 150. 119. 79.8
#19 D 13.1 119. 143. 88.9
#20 D 10.3 139. 117. 76.5
关于r - 按组 pmap 列以获得加权行均值，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/66259932/