![](/img/trans.png)
[英]R dplyr: how to use ... with summarize(across()) when ... will refer to a variable name within the data?
[英]refer to column name from variable in across in dplyr
給定參考列z
,我想使用dplyr
將每一列轉換為:
x = log(x) - log(z)
我希望z
是一個字符串,或者更好的是,一個帶引號的表達式(例如,由用戶輸入 - 所有這些都在一個函數內)。
這是我嘗試過的:
library(dplyr)
m <- data.frame(x=1:5,y=11:15,z=21:25)
denom = "z"
這有效:
m %>%
mutate(across(x:z ,
list(~ log(.) - log(z) )))
這失敗了:
m %>%
mutate(across(x:z ,
list(~ log(.) - log(rlang::sym(denom)))))
# Error: Problem with `mutate()` input `..1`.
# ℹ `..1 = across(x:z, list(~log(.) - log(rlang::sym(denom))))`.
# ✖ non-numeric argument to mathematical function
# Run `rlang::last_error()` to see where the error occurred.
這也失敗了:
m %>%
mutate(across(x:z ,
list(~ log(.) - log(!!denom) )))
# Error: Problem with `mutate()` input `..1`.
# ℹ `..1 = across(x:z, list(~log(.) - log("z")))`.
# ✖ non-numeric argument to mathematical function
# Run `rlang::last_error()` to see where the error occurred.
# > #list(~ log(.) - log(rlang::sym(denom)))))
您還可以使用get()
:
m %>% mutate(across(.fns = list(~ log(.) - log(get(denom)))))
使用rlang
中的.data
代詞,您可以:
library(dplyr)
m <- data.frame(x = 1:5, y = 11:15, z = 21:25)
denom <- "z"
m %>% mutate(across(
x:z,
list(~ log(.) - log(.data[[denom]]))
))
#> x y z x_1 y_1 z_1
#> 1 1 11 21 -3.044522 -0.6466272 0
#> 2 2 12 22 -2.397895 -0.6061358 0
#> 3 3 13 23 -2.036882 -0.5705449 0
#> 4 4 14 24 -1.791759 -0.5389965 0
#> 5 5 15 25 -1.609438 -0.5108256 0
我不知道,這是否是一種很好的編碼方式,但你可以這樣做
library(dplyr)
m %>%
mutate(across(x:z ,
list(~ log(.) - log(!!as.name(denom)) )))
如果存在比簡單選擇一列更復雜的情況,可以使用str2lang
(或parse
) eval
。 如果它是一個表達式,它可以直接在eval
denom <- "z"
m %>% mutate(across(x:z, list(~ log(.) - log(eval(str2lang(denom))) )))
# x y z x_1 y_1 z_1
#1 1 11 21 -3.044522 -0.6466272 0
#2 2 12 22 -2.397895 -0.6061358 0
#3 3 13 23 -2.036882 -0.5705449 0
#4 4 14 24 -1.791759 -0.5389965 0
#5 5 15 25 -1.609438 -0.5108256 0
denom <- expression(z)
m %>% mutate(across(x:z, list(~ log(.) - log(eval(denom)) )))
# x y z x_1 y_1 z_1
#1 1 11 21 -3.044522 -0.6466272 0
#2 2 12 22 -2.397895 -0.6061358 0
#3 3 13 23 -2.036882 -0.5705449 0
#4 4 14 24 -1.791759 -0.5389965 0
#5 5 15 25 -1.609438 -0.5108256 0
m %>% mutate(across(x:z, list(~ log(.) - log(z) )))
# x y z x_1 y_1 z_1
#1 1 11 21 -3.044522 -0.6466272 0
#2 2 12 22 -2.397895 -0.6061358 0
#3 3 13 23 -2.036882 -0.5705449 0
#4 4 14 24 -1.791759 -0.5389965 0
#5 5 15 25 -1.609438 -0.5108256 0
更復雜:
denom <- "x + y"
m %>% mutate(across(x:z, list(~ log(.) - log(eval(str2lang(denom))) )))
# x y z x_1 y_1 z_1
#1 1 11 21 -2.484907 -0.08701138 0.5596158
#2 2 12 22 -1.945910 -0.15415068 0.4519851
#3 3 13 23 -1.673976 -0.20763936 0.3629055
#4 4 14 24 -1.504077 -0.25131443 0.2876821
#5 5 15 25 -1.386294 -0.28768207 0.2231436
denom <- expression(x + y)
m %>% mutate(across(x:z, list(~ log(.) - log(eval(denom)) )))
# x y z x_1 y_1 z_1
#1 1 11 21 -2.484907 -0.08701138 0.5596158
#2 2 12 22 -1.945910 -0.15415068 0.4519851
#3 3 13 23 -1.673976 -0.20763936 0.3629055
#4 4 14 24 -1.504077 -0.25131443 0.2876821
#5 5 15 25 -1.386294 -0.28768207 0.2231436
m %>% mutate(across(x:z, list(~ log(.) - log(x + y) )))
# x y z x_1 y_1 z_1
#1 1 11 21 -2.484907 -0.08701138 0.5596158
#2 2 12 22 -1.945910 -0.15415068 0.4519851
#3 3 13 23 -1.673976 -0.20763936 0.3629055
#4 4 14 24 -1.504077 -0.25131443 0.2876821
#5 5 15 25 -1.386294 -0.28768207 0.2231436
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.