问题描述
df <- tibble::rownames_to_column(USArrests,"State") %>%
tidyr::pivot_longer(cols = -State)
head(df)
# A tibble: 6 x 3
State name value
<chr> <chr> <dbl>
1 Alabama Murder 13.2
2 Alabama Assault 236
3 Alabama UrbanPop 58
4 Alabama Rape 21.2
5 Alaska Murder 10
6 Alaska Assault 263
在单独的列表对象 l
中,我有列,我需要从数据框中删除这些列。元素名称是列名称,值对应于我要删除的行:
l <- list(State = c("Alabama","Pennsylvania","Texas"),name = c("Murder","Assault"))
硬编码它会这样做:
dplyr::filter(df,!State %in% c("Alabama",!name %in% c("Murder","Assault"))
State name value
<chr> <chr> <dbl>
1 Alaska UrbanPop 48
2 Alaska Rape 44.5
3 Arizona UrbanPop 80
4 Arizona Rape 31
5 Arkansas UrbanPop 50
6 Arkansas Rape 19.5
7 California UrbanPop 91
8 California Rape 40.6
9 Colorado UrbanPop 78
10 Colorado Rape 38.7
# ... with 84 more rows
但是,l
经常更改,因此我不能/不想进行硬编码。我尝试了以下操作,但只计算最后一个表达式:
library(purrr)
filter_expr <- imap_chr(l,~ paste0("! ",.y," %in% c(\"",paste(.x,collapse = "\",\""),"\")")) %>% parse(text = .)
filter(df,eval(filter_expr))
State name value
<chr> <chr> <dbl>
1 Alabama UrbanPop 58
2 Alabama Rape 21.2
3 Alaska UrbanPop 48
4 Alaska Rape 44.5
5 Arizona UrbanPop 80
6 Arizona Rape 31
7 Arkansas UrbanPop 50
8 Arkansas Rape 19.5
9 California UrbanPop 91
10 California Rape 40.6
# ... with 90 more rows
当过滤条件存储在像 df
这样更符合 tidyverse 习惯的结构中时,有没有办法过滤 l
?
我考虑过这个 SO answer,但是,表达式不是动态的。
解决方法
我们可以在 across
中使用 filter
来循环 'l' 的 names
,通过使用列名({{1 }}) 和否定 (cur_column()
)。请注意,!
目前仅适用于 cur_column()
而不适用于 across
(if_all/if_any
-dplyr
在 1.0.6
)
R 4.1.0
-输出
library(dplyr)
df %>%
filter(across(all_of(names(l)),~ !. %in% l[[cur_column()]]))
如果我们可以设置一个属性,我们可以使用 # A tibble: 94 x 3
# State name value
# <chr> <chr> <dbl>
# 1 Alaska UrbanPop 48
# 2 Alaska Rape 44.5
# 3 Arizona UrbanPop 80
# 4 Arizona Rape 31
# 5 Arkansas UrbanPop 50
# 6 Arkansas Rape 19.5
# 7 California UrbanPop 91
# 8 California Rape 40.6
# 9 Colorado UrbanPop 78
#10 Colorado Rape 38.7
# … with 84 more rows
if_all
或者用library(magrittr)
df %>%
mutate(across(all_of(names(l)),~ set_attr(.,'cn',cur_column()))) %>%
filter(if_all(all_of(names(l)),~ ! . %in% l[[attr(.,'cn')]]))
imap/reduce
或者另一个选项是library(purrr)
df %>%
filter(imap(l,~ !cur_data()[[.y]] %in% .x) %>%
reduce(`&`))
anti_join
,
这里的另一个潜在选择是使用 purrr
创建一个逻辑向量,该向量允许 &
与 |
条件,并且可以访问当前列名 (.y
) 而无需cur_column
,只能在 across
内使用:
df %>%
filter(imap(l,~ !df[[.y]] %in% .x) %>% reduce(`&`)) # can use magrittr::and
输出
State name value
<chr> <chr> <dbl>
1 Alaska UrbanPop 48
2 Alaska Rape 44.5
3 Arizona UrbanPop 80
4 Arizona Rape 31
5 Arkansas UrbanPop 50
6 Arkansas Rape 19.5
7 California UrbanPop 91
8 California Rape 40.6
9 Colorado UrbanPop 78
10 Colorado Rape 38.7
# ... with 84 more rows
或变体是:
df %>%
filter(imap(l,~ !df[[.y]] %in% .x) %>% reduce(`|`)) # can use magrittr::or