高阶函数 （higher-order functions）

泛函（functionals）是将函数作为输入并返回向量作为输出的函数。

在R语言中，purrr是一个函数式编程包，它提供了一种简洁的方式来迭代数据结构。这个包的名字来源于它所提供的核心功能，即“纯”（pure）函数的迭代。在函数式编程中，纯函数是指给定相同的输入总是返回相同输出的函数，并且没有副作用（如更改全局变量的值->、写入磁盘write_csv()或显示在屏幕上print()）。

Show the code

10.1 map 变体

参数 \ 返回 列表 原子 同类型 无
一个参数 map() map_lgl(), … modify() walk()
两个参数 map2() map2_lgl(), … modify2() walk2()
一个参数 + 索引 imap() imap_lgl(), … imodify() iwalk()
N 个参数 pmap() pmap_lgl(), … — pwalk()

map 的基本构造如下，实际上是用 C 语言编写的，以节省性能，保留名称，并支持一些快捷方式。

Show the code
simple_map <- function(x, f, ...) {
  out <- vector("list", length(x)) # 分配一个与输入长度相同的列表
  for (i in seq_along(x)) {        # 用 for 循环填充列表
    out[[i]] <- f(x[[i]], ...)
  }
  out
}

10.1.1 返回列表

purrr::map(1:3,f) 接受一个向量和一个函数，为向量的每个元素调用一次函数，并在列表中返回结果，等价于list(f(1), f(2), f(3))

Show the code
triple <- function(x) x * 3
map(1:3, triple)
#> [[1]]
#> [1] 3
#> 
#> [[2]]
#> [1] 6
#> 
#> [[3]]
#> [1] 9

10.1.2 返回原子向量

map_lgl()、map_int()、map_dbl()、map_chr() 每个都返回一个指定类型的原子向量

Show the code
map_chr(mtcars, typeof) 
#>      mpg      cyl     disp       hp     drat       wt     qsec       vs 
#> "double" "double" "double" "double" "double" "double" "double" "double" 
#>       am     gear     carb 
#> "double" "double" "double"
map_lgl(mtcars, is.double)  
#>  mpg  cyl disp   hp drat   wt qsec   vs   am gear carb 
#> TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE

n_unique <- function(x)length(unique(x)) 
map_int(mtcars, n_unique)  
#>  mpg  cyl disp   hp drat   wt qsec   vs   am gear carb 
#>   25    3   27   22   22   29   30    2    2    3    6

map_dbl(mtcars, mean)
#>        mpg        cyl       disp         hp       drat         wt       qsec 
#>  20.090625   6.187500 230.721875 146.687500   3.596563   3.217250  17.848750 
#>         vs         am       gear       carb 
#>   0.437500   0.406250   3.687500   2.812500
Show the code
by_cyl <- split(mtcars, mtcars$cyl)
by_cyl %>% 
  map(~ lm(mpg ~ wt, data = .x)) %>% 
  map(coef) %>% 
  map_dbl(2)
#>         4         6         8 
#> -5.647025 -2.780106 -2.192438

10.1.3 返回与输入类型相同的输出：modify()

Show the code
df <- data.frame(
  x = 1:3,
  y = 6:4
)

# 返回列表
map(df, ~ .x * 2)
#> $x
#> [1] 2 4 6
#> 
#> $y
#> [1] 12 10  8

# 返回数据框
modify(df, ~ .x * 2)
x y
2 12
4 10
6 8

10.1.4 无返回

某些函数主要是为了它们的副作用（如cat()、write_csv()、ggsave()）而被调用，并且捕获和存储它们的结果是没有意义的。

Show the code
welcome <- function(x) {
  cat("Welcome ", x, "!\n", sep = "")
}
names <- c("Hadley", "Jenny")
map(names, welcome)
#> Welcome Hadley!
#> Welcome Jenny!
#> [[1]]
#> NULL
#> 
#> [[2]]
#> NULL
Show the code
walk(names, welcome)
#> Welcome Hadley!
#> Welcome Jenny!
Show the code
cyls <- split(mtcars, mtcars$cyl)
paths <- paste0("data/cyl-", names(cyls), ".csv")
walk2(cyls, paths, write.csv)

等价于write.csv(cyls[[1]]paths[[1]])、write.csv(cyls[[2]],paths[[2]])

write.csv(cyls[[3]], paths[[3]])

10.1.5 两个输入

Show the code
set.seed(10)
xs <- map(1:8, ~ runif(10))


# 权重
set.seed(100)
ws <- map(1:8, ~ rpois(10, 5) + 1)

xs[[1]][[1]] <- NA
xs
#> [[1]]
#>  [1]         NA 0.30676851 0.42690767 0.69310208 0.08513597 0.22543662
#>  [7] 0.27453052 0.27230507 0.61582931 0.42967153
#> 
#> [[2]]
#>  [1] 0.65165567 0.56773775 0.11350898 0.59592531 0.35804998 0.42880942
#>  [7] 0.05190332 0.26417767 0.39879073 0.83613414
#> 
#> [[3]]
#>  [1] 0.8647212 0.6153524 0.7751099 0.3555687 0.4058500 0.7066469 0.8382877
#>  [8] 0.2395891 0.7707715 0.3558977
#> 
#> [[4]]
#>  [1] 0.53559704 0.09308813 0.16980304 0.89983245 0.42263761 0.74774647
#>  [7] 0.82265258 0.95465365 0.68544451 0.50050323
#> 
#> [[5]]
#>  [1] 0.27548386 0.22890394 0.01443391 0.72896456 0.24988047 0.16118328
#>  [7] 0.01704265 0.48610035 0.10290017 0.80154700
#> 
#> [[6]]
#>  [1] 0.3543281 0.9364325 0.2458664 0.4731415 0.1915609 0.5832220 0.4594732
#>  [8] 0.4674340 0.3998326 0.5052856
#> 
#> [[7]]
#>  [1] 0.03188816 0.11446759 0.46893548 0.39698674 0.83361919 0.76112174
#>  [7] 0.57335645 0.44750805 0.08380201 0.21913855
#> 
#> [[8]]
#>  [1] 0.07557029 0.53442678 0.64135658 0.52573932 0.03928139 0.54585984
#>  [7] 0.37276310 0.96130241 0.25734157 0.20795168

# 未加权均值
map_dbl(xs, mean)
#> [1]        NA 0.4266693 0.5927795 0.5831959 0.3066440 0.4616577 0.3930824
#> [8] 0.4161593
# 加权均值  weighted.mean()
map2_dbl(xs, ws, weighted.mean)
#> [1]        NA 0.4590300 0.5834602 0.5998278 0.3067582 0.4448242 0.3947274
#> [8] 0.4418898
map2_dbl(xs, ws, weighted.mean, na.rm = TRUE)
#> [1] 0.3479728 0.4590300 0.5834602 0.5998278 0.3067582 0.4448242 0.3947274
#> [8] 0.4418898

10.1.6 遍历值和索引

imap(x, f)等价于map2(x, names(x), f)、map2(x, seq_along(x), f)

imap()通常可用于构建标签：

Show the code
imap_chr(iris, ~ paste0("The first value of ", .y, " is ", .x[[1]]))
#>                             Sepal.Length 
#> "The first value of Sepal.Length is 5.1" 
#>                              Sepal.Width 
#>  "The first value of Sepal.Width is 3.5" 
#>                             Petal.Length 
#> "The first value of Petal.Length is 1.4" 
#>                              Petal.Width 
#>  "The first value of Petal.Width is 0.2" 
#>                                  Species 
#>   "The first value of Species is setosa"
map2_chr(iris,names(iris),~ paste0("The first value of ", .y, " is ", .x[[1]]))
#>                             Sepal.Length 
#> "The first value of Sepal.Length is 5.1" 
#>                              Sepal.Width 
#>  "The first value of Sepal.Width is 3.5" 
#>                             Petal.Length 
#> "The first value of Petal.Length is 1.4" 
#>                              Petal.Width 
#>  "The first value of Petal.Width is 0.2" 
#>                                  Species 
#>   "The first value of Species is setosa"

如果向量未命名，则第二个参数将是索引：

Show the code
x <- map(1:6, ~ sample(1000, 10))
imap_chr(x, ~ paste0("The maximum value of ", .y, " is ", max(.x)))
#> [1] "The maximum value of 1 is 965" "The maximum value of 2 is 978"
#> [3] "The maximum value of 3 is 949" "The maximum value of 4 is 871"
#> [5] "The maximum value of 5 is 843" "The maximum value of 6 is 848"
map2_chr(x,seq_along(x), ~ paste0("The highest value of ", .y, " is ", max(.x)))
#> [1] "The highest value of 1 is 965" "The highest value of 2 is 978"
#> [3] "The highest value of 3 is 949" "The highest value of 4 is 871"
#> [5] "The highest value of 5 is 843" "The highest value of 6 is 848"

10.1.7 任意数量的输入

map2(x, y, f ) 等价于 pmap(list(x, y), f)

Show the code
pmap_dbl(list(xs, ws), weighted.mean)
#> [1]        NA 0.4590300 0.5834602 0.5998278 0.3067582 0.4448242 0.3947274
#> [8] 0.4418898
pmap_dbl(list(xs, ws), weighted.mean, na.rm = TRUE)
#> [1] 0.3479728 0.4590300 0.5834602 0.5998278 0.3067582 0.4448242 0.3947274
#> [8] 0.4418898
Show the code
# 命名列表 调用函数
params <- tibble::tribble(
  ~ n, ~ min, ~ max,
   1L,     0,     1,
   2L,    10,   100,
   3L,   100,  1000
)
typeof(params)
#> [1] "list"
pmap(params, runif)
#> [[1]]
#> [1] 0.08703467
#> 
#> [[2]]
#> [1] 22.92254 91.76306
#> 
#> [[3]]
#> [1] 210.2793 755.9765 955.3395

10.2 reduce 家族

通过迭代应用两个输入的函数（a binary function）将列表简化为单个值。

10.2.1 reduce()

reduce()获取长度为 n 的向量，并通过一次调用具有一对值的函数来生成长度为 1 的向量

reduce(1:4, f)等价于f(f(f(1, 2), 3), 4)

Show the code
l <- map(1:4, ~ sample(1:10, 15, replace = T))

# 交集
reduce(l, intersect)
#> [1] 10  5  7  4

# 并集
reduce(l, union)
#>  [1] 10  5  7  8  1  6  3  4  2  9
Show the code
simple_reduce <- function(x, f) {
  out <- x[[1]]
  for (i in seq(2, length(x))) {
    out <- f(out, x[[i]])
  }
  out
}

10.2.2 accumulate()

第一个变体 ，不仅返回最终结果，还返回所有中间结果

Show the code
accumulate(l, intersect)
#> [[1]]
#>  [1] 10  5  7  8  1  8  6 10  3  6 10  6  8  3  4
#> 
#> [[2]]
#> [1] 10  5  7  1  4
#> 
#> [[3]]
#> [1] 10  5  7  1  4
#> 
#> [[4]]
#> [1] 10  5  7  4
Show the code
x <- c(4, 3, 10)
reduce(x, `+`)
#> [1] 17

accumulate(x, `+`)
#> [1]  4  7 17

10.2.3 reduce2(x,y,f,.init)

10.3 谓词泛函

在编程和数学中，“谓词”（Predicate）是一个表达式，它返回一个布尔值（真或假）。谓词通常用于逻辑和函数式编程中，作为判断或决策的基础。

谓词泛函（Predicate Functionals）是谓词的一个特例，它是一个接受一个或多个参数，并返回一个布尔值的函数。谓词泛函在很多编程语言中都很常见，尤其是在支持函数式编程的语言中。

谓词泛函的一些关键特点包括：

  1. 返回布尔值：谓词泛函的输出是一个布尔值，即真（true）或假（false）。

  2. 接受参数：谓词泛函可以接收一个或多个参数，这些参数的值会影响函数的返回结果。

  3. 无副作用：在函数式编程中，谓词泛函通常是纯函数，即它们不会产生副作用，也不会修改外部状态。

  4. 逻辑判断：谓词泛函常用于逻辑判断，如条件语句（if）、循环（while、for）等。

Show the code
df <- data.frame(
  num1 = c(0, 10, 20),
  num2 = c(5, 6, 7),
  chr1 = c("a", "b", "c")
)
detect(df, is.factor)
#> NULL
detect_index(df, is.factor)
#> [1] 0
str(keep(df, is.factor))
#> 'data.frame':    3 obs. of  0 variables
str(discard(df, is.factor))
#> 'data.frame':    3 obs. of  3 variables:
#>  $ num1: num  0 10 20
#>  $ num2: num  5 6 7
#>  $ chr1: chr  "a" "b" "c"

map变体也有谓词泛函

Show the code
str(map_if(df, is.numeric, mean))
#> List of 3
#>  $ num1: num 10
#>  $ num2: num 6
#>  $ chr1: chr [1:3] "a" "b" "c"
str(modify_if(df, is.numeric, mean))
#> 'data.frame':    3 obs. of  3 variables:
#>  $ num1: num  10 10 10
#>  $ num2: num  6 6 6
#>  $ chr1: chr  "a" "b" "c"
str(map(keep(df, is.numeric), mean))
#> List of 2
#>  $ num1: num 10
#>  $ num2: num 6

10.4 base 包泛函

10.4.1 apply()

base::apply()

  • X，要汇总的矩阵或数组。

  • MARGIN，一个整数向量，给出要汇总的维度， 1 = 行，2 = 列，以此类推（参数名称来自思考 联合分布的边距）。

  • FUN，一个汇总函数。

Show the code
a3d <- array(1:24, c(4, 3, 2))
a3d
#> , , 1
#> 
#>      [,1] [,2] [,3]
#> [1,]    1    5    9
#> [2,]    2    6   10
#> [3,]    3    7   11
#> [4,]    4    8   12
#> 
#> , , 2
#> 
#>      [,1] [,2] [,3]
#> [1,]   13   17   21
#> [2,]   14   18   22
#> [3,]   15   19   23
#> [4,]   16   20   24
apply(a3d, c(1, 2), mean)
#>      [,1] [,2] [,3]
#> [1,]    7   11   15
#> [2,]    8   12   16
#> [3,]    9   13   17
#> [4,]   10   14   18

lapply()、sapply() 和 vapply() 针对的都是列表结构的数据，sapply()是简化（simplified）版本的 lapply()，而 vapply() 则在 sapply() 的基础上加了结果验证，以保证可靠性。

Show the code
set.seed(10)

temp <- list(
  35 + rnorm(10, mean = 1, sd = 10),
  20 + rnorm(5, mean = 1, sd = 3),
  25 + rnorm(22, mean = 2, sd = 6),
  33 + rnorm(14, mean = 4, sd = 20)
)
basic <- function(x) {
  c(min = min(x), mean = mean(x), median = median(x), max = max(x))
}
Show the code
lapply(temp, basic)
#> [[1]]
#>      min     mean   median      max 
#> 19.73327 31.09343 32.89923 39.89794 
#> 
#> [[2]]
#>      min     mean   median      max 
#> 20.28530 23.00890 23.26734 24.30534 
#> 
#> [[3]]
#>      min     mean   median      max 
#> 13.88828 23.72360 24.09008 32.81140 
#> 
#> [[4]]
#>       min      mean    median       max 
#>  1.818265 31.437784 29.748184 58.731028
sapply(temp, basic)
#>            [,1]     [,2]     [,3]      [,4]
#> min    19.73327 20.28530 13.88828  1.818265
#> mean   31.09343 23.00890 23.72360 31.437784
#> median 32.89923 23.26734 24.09008 29.748184
#> max    39.89794 24.30534 32.81140 58.731028
Show the code
vapply(temp, basic, numeric(4))
#>            [,1]     [,2]     [,3]      [,4]
#> min    19.73327 20.28530 13.88828  1.818265
#> mean   31.09343 23.00890 23.72360 31.437784
#> median 32.89923 23.26734 24.09008 29.748184
#> max    39.89794 24.30534 32.81140 58.731028

vapply() 第 3 个参数传入对每一个子集调用函数后结果的预期，上述设定为包含 4 个元素的数值型向量。

10.5 apply函数簇

10.5.1 apply()

Show the code
apply(X = mtcars, MARGIN = 2, FUN = mean)
#>        mpg        cyl       disp         hp       drat         wt       qsec 
#>  20.090625   6.187500 230.721875 146.687500   3.596563   3.217250  17.848750 
#>         vs         am       gear       carb 
#>   0.437500   0.406250   3.687500   2.812500

10.5.2 aggregate()

Show the code
aggregate(x = mtcars,
          by = list(am=mtcars$am,cyl=mtcars$cyl),
          FUN = mean)
am cyl mpg cyl disp hp drat wt qsec vs am gear carb
0 4 22.90000 4 135.8667 84.66667 3.770000 2.935000 20.97000 1.000 0 3.666667 1.666667
1 4 28.07500 4 93.6125 81.87500 4.183750 2.042250 18.45000 0.875 1 4.250000 1.500000
0 6 19.12500 6 204.5500 115.25000 3.420000 3.388750 19.21500 1.000 0 3.500000 2.500000
1 6 20.56667 6 155.0000 131.66667 3.806667 2.755000 16.32667 0.000 1 4.333333 4.666667
0 8 15.05000 8 357.6167 194.16667 3.120833 4.104083 17.14250 0.000 0 3.000000 3.083333
1 8 15.40000 8 326.0000 299.50000 3.880000 3.370000 14.55000 0.000 1 5.000000 6.000000
Show the code
aggregate(.~am+cyl,
          data = mtcars,
          FUN = mean)
am cyl mpg disp hp drat wt qsec vs gear carb
0 4 22.90000 135.8667 84.66667 3.770000 2.935000 20.97000 1.000 3.666667 1.666667
1 4 28.07500 93.6125 81.87500 4.183750 2.042250 18.45000 0.875 4.250000 1.500000
0 6 19.12500 204.5500 115.25000 3.420000 3.388750 19.21500 1.000 3.500000 2.500000
1 6 20.56667 155.0000 131.66667 3.806667 2.755000 16.32667 0.000 4.333333 4.666667
0 8 15.05000 357.6167 194.16667 3.120833 4.104083 17.14250 0.000 3.000000 3.083333
1 8 15.40000 326.0000 299.50000 3.880000 3.370000 14.55000 0.000 5.000000 6.000000
Show the code
aggregate(mpg~am+cyl,
          data = mtcars,
          FUN = summary)
am cyl mpg
0 4 21.50000
1 4 21.40000
0 6 17.80000
1 6 19.70000
0 8 10.40000
1 8 15.00000

10.5.3 by(), tapply()

by()是应用于数据框的tapply的面向对象包装器。

Show the code
by(data = mtcars[,1:3],
   INDICES = list(cyl=mtcars$cyl),
   FUN = summary)
#> cyl: 4
#>       mpg             cyl         disp       
#>  Min.   :21.40   Min.   :4   Min.   : 71.10  
#>  1st Qu.:22.80   1st Qu.:4   1st Qu.: 78.85  
#>  Median :26.00   Median :4   Median :108.00  
#>  Mean   :26.66   Mean   :4   Mean   :105.14  
#>  3rd Qu.:30.40   3rd Qu.:4   3rd Qu.:120.65  
#>  Max.   :33.90   Max.   :4   Max.   :146.70  
#> ------------------------------------------------------------ 
#> cyl: 6
#>       mpg             cyl         disp      
#>  Min.   :17.80   Min.   :6   Min.   :145.0  
#>  1st Qu.:18.65   1st Qu.:6   1st Qu.:160.0  
#>  Median :19.70   Median :6   Median :167.6  
#>  Mean   :19.74   Mean   :6   Mean   :183.3  
#>  3rd Qu.:21.00   3rd Qu.:6   3rd Qu.:196.3  
#>  Max.   :21.40   Max.   :6   Max.   :258.0  
#> ------------------------------------------------------------ 
#> cyl: 8
#>       mpg             cyl         disp      
#>  Min.   :10.40   Min.   :8   Min.   :275.8  
#>  1st Qu.:14.40   1st Qu.:8   1st Qu.:301.8  
#>  Median :15.20   Median :8   Median :350.5  
#>  Mean   :15.10   Mean   :8   Mean   :353.1  
#>  3rd Qu.:16.25   3rd Qu.:8   3rd Qu.:390.0  
#>  Max.   :19.20   Max.   :8   Max.   :472.0

by(data = mtcars[,1:3],
   INDICES = factor(mtcars$cyl),
   FUN = function(x) lm( disp ~ mpg, data = x))
#> factor(mtcars$cyl): 4
#> 
#> Call:
#> lm(formula = disp ~ mpg, data = x)
#> 
#> Coefficients:
#> (Intercept)          mpg  
#>     233.067       -4.798  
#> 
#> ------------------------------------------------------------ 
#> factor(mtcars$cyl): 6
#> 
#> Call:
#> lm(formula = disp ~ mpg, data = x)
#> 
#> Coefficients:
#> (Intercept)          mpg  
#>     125.122        2.947  
#> 
#> ------------------------------------------------------------ 
#> factor(mtcars$cyl): 8
#> 
#> Call:
#> lm(formula = disp ~ mpg, data = x)
#> 
#> Coefficients:
#> (Intercept)          mpg  
#>      560.87       -13.76

tapply(X = mtcars[,1:3],
       INDEX = list(cyl=mtcars$cyl),
       FUN = summary)
#> $`4`
#>       mpg             cyl         disp       
#>  Min.   :21.40   Min.   :4   Min.   : 71.10  
#>  1st Qu.:22.80   1st Qu.:4   1st Qu.: 78.85  
#>  Median :26.00   Median :4   Median :108.00  
#>  Mean   :26.66   Mean   :4   Mean   :105.14  
#>  3rd Qu.:30.40   3rd Qu.:4   3rd Qu.:120.65  
#>  Max.   :33.90   Max.   :4   Max.   :146.70  
#> 
#> $`6`
#>       mpg             cyl         disp      
#>  Min.   :17.80   Min.   :6   Min.   :145.0  
#>  1st Qu.:18.65   1st Qu.:6   1st Qu.:160.0  
#>  Median :19.70   Median :6   Median :167.6  
#>  Mean   :19.74   Mean   :6   Mean   :183.3  
#>  3rd Qu.:21.00   3rd Qu.:6   3rd Qu.:196.3  
#>  Max.   :21.40   Max.   :6   Max.   :258.0  
#> 
#> $`8`
#>       mpg             cyl         disp      
#>  Min.   :10.40   Min.   :8   Min.   :275.8  
#>  1st Qu.:14.40   1st Qu.:8   1st Qu.:301.8  
#>  Median :15.20   Median :8   Median :350.5  
#>  Mean   :15.10   Mean   :8   Mean   :353.1  
#>  3rd Qu.:16.25   3rd Qu.:8   3rd Qu.:390.0  
#>  Max.   :19.20   Max.   :8   Max.   :472.0

10.5.4 lapply(), sapply()

lapply() returns a list of the same length as X,

sapply() is a user-friendly version and wrapper of lapply() by default returning a vector or matrix

Show the code
x <- list(a = 1:10, beta = exp(-3:3), logic = c(TRUE,FALSE,FALSE,TRUE))
lapply(X = x,FUN = quantile)
#> $a
#>    0%   25%   50%   75%  100% 
#>  1.00  3.25  5.50  7.75 10.00 
#> 
#> $beta
#>          0%         25%         50%         75%        100% 
#>  0.04978707  0.25160736  1.00000000  5.05366896 20.08553692 
#> 
#> $logic
#>   0%  25%  50%  75% 100% 
#>  0.0  0.0  0.5  1.0  1.0

sapply(x, quantile)
#>          a        beta logic
#> 0%    1.00  0.04978707   0.0
#> 25%   3.25  0.25160736   0.0
#> 50%   5.50  1.00000000   0.5
#> 75%   7.75  5.05366896   1.0
#> 100% 10.00 20.08553692   1.0

10.5.5 vapply

Show the code
by_cyl <- split(mtcars, mtcars$cyl)
models <- lapply(by_cyl, function(data) lm(mpg ~ wt, data = data))
models
#> $`4`
#> 
#> Call:
#> lm(formula = mpg ~ wt, data = data)
#> 
#> Coefficients:
#> (Intercept)           wt  
#>      39.571       -5.647  
#> 
#> 
#> $`6`
#> 
#> Call:
#> lm(formula = mpg ~ wt, data = data)
#> 
#> Coefficients:
#> (Intercept)           wt  
#>       28.41        -2.78  
#> 
#> 
#> $`8`
#> 
#> Call:
#> lm(formula = mpg ~ wt, data = data)
#> 
#> Coefficients:
#> (Intercept)           wt  
#>      23.868       -2.192
vapply(models, function(x) coef(x)[[2]], double(1))
#>         4         6         8 
#> -5.647025 -2.780106 -2.192438

10.5.6 数学泛函