将变量传递到MAP功能公式中的.x $

Question

第一个示例有效，但是我想沿示例2的行中移动更多的重复代码。我还没有找到正确的rlang方法来处理.x $ x。

library(tidyverse)

# This produces the desired result:
slope <- function(y, x){
  coef(lm(y ~ x))[2]
}

tribble(
  ~group, ~y, ~x1, ~x2,
  "a", 1, 1, 4,
  "a", 2, 2, 5,
  "a", 3, 3, 6,
  "b", 1, 4, 8,
  "b", 2, 6, 12,
  "b", 3, 8, 16
) |> 
  nest(data = -group) |>
  mutate(
    slope1 = map_dbl(data, ~ slope(.x$y, .x$x1)),
    slope2 = map_dbl(data, ~ slope(.x$y, .x$x2))
  )
#> # A tibble: 2 × 4
#>   group data             slope1 slope2
#>   <chr> <list>            <dbl>  <dbl>
#> 1 a     <tibble [3 × 3]>    1     1   
#> 2 b     <tibble [3 × 3]>    0.5   0.25

# I would like to use rlang to further simplify with something like this,
# but I can't see how to combine the variable with .x$:
slope <- function(x) {
  map_dbl(data, ~ coef(lm(.x$y ~ .x$x))[2])
}

tribble(
  ~group, ~y, ~x1, ~x2,
  "a", 1, 1, 4,
  "a", 2, 2, 5,
  "a", 3, 3, 6,
  "b", 1, 4, 8,
  "b", 2, 6, 12,
  "b", 3, 8, 16
) |> 
  nest(data = -group) |>
  mutate(
    slope1 = slope(x1),
    slope2 = slope(x2)
  )
#> Error in `mutate()`:
#> ! Problem while computing `slope1 = slope(x1)`.
#> Caused by error in `purrr:::stop_bad_type()`:
#> ! `.x` must be a vector, not a function

^由

Answer 1

要使第二版中的代码工作，您的slope函数将需要获取数据参数以及您希望回归的列的名称。然后，列名称需要将其制成lm调用的公式，可以通过几种方式完成，也许最容易使用一些字符串解析和as.formula。为简单起见，我还将使用data lm的参数。

slope <- function(data, var) {
  
  f <- as.formula(paste('y', deparse(substitute(var)), sep = "~"))
  map_dbl(data, ~ coef(lm(f, data = .x))[2])
}

因此，您有：

tribble(
  ~group, ~y, ~x1, ~x2,
  "a", 1, 1, 4,
  "a", 2, 2, 5,
  "a", 3, 3, 6,
  "b", 1, 4, 8,
  "b", 2, 6, 12,
  "b", 3, 8, 16
) |> 
  nest(data = -group) |>
  mutate(
    slope1 = slope(data, x1),
    slope2 = slope(data, x2)
  )

#> # A tibble: 2 x 4
#>   group data             slope1 slope2
#>   <chr> <list>            <dbl>  <dbl>
#> 1 a     <tibble [3 x 3]>    1     1   
#> 2 b     <tibble [3 x 3]>    0.5   0.25

我个人认为，这样做的一种更强大，更通用的方法是传递一个公式：

slope <- function(data, f) map_dbl(data, ~ coef(lm(f, data = .x))[2])

tribble(
  ~group, ~y, ~x1, ~x2,
  "a", 1, 1, 4,
  "a", 2, 2, 5,
  "a", 3, 3, 6,
  "b", 1, 4, 8,
  "b", 2, 6, 12,
  "b", 3, 8, 16
) |> 
  nest(data = -group) |>
  mutate(
    slope1 = slope(data, y ~ x1),
    slope2 = slope(data, y ~ x2)
  )

#> # A tibble: 2 x 4
#>   group data             slope1 slope2
#>   <chr> <list>            <dbl>  <dbl>
#> 1 a     <tibble [3 x 3]>    1     1   
#> 2 b     <tibble [3 x 3]>    0.5   0.25