16

数据规整进阶

嵌套与高级变换

Tidyverse 篇

接着上一章，罗列一些tidyr的函数

R 
library(tidyverse)

fill() 缺失值填充

利用所在列的上下值进行缺失值填充

R 
sales <- tibble::tribble(
  ~quarter, ~year, ~sales,
      "Q1",  2000,  66013,
      "Q2",    NA,  69182,
      "Q3",    NA,  53175,
      "Q4",    NA,  21001,
      "Q1",  2001,  46036,
      "Q2",    NA,  58842,
      "Q3",    NA,  44568,
      "Q4",    NA,  50197,
      "Q1",  2002,  39113,
      "Q2",    NA,  41668,
      "Q3",    NA,  30144,
      "Q4",    NA,  52897
  )
sales
R 
sales %>% fill(year)

也可以控制填充的方向

R 
sales %>% fill(year, .direction = "up")

expand() 与 complete()

指定数据框的若干列，根据其向量元素值，产生所有可能的交叉组合

R 
df <- tibble::tribble(
  ~x, ~y, ~z,
  1L, 1L, 4L,
  1L, 2L, 5L,
  2L, 1L, NA,
  3L, 2L, 6L
)


df %>% expand(x, y)

nesting()用于限定只产生数据框已出现的组合。

R 
df %>% expand(nesting(x, y))
R 
df %>% expand(nesting(x, y), z)

complete() 补全，可以看做是 expand(nesting()) + fill()

R 
df %>% complete(x, y)
R 
df %>% complete(x, y, fill = list(z = 0))

数据在complete补全的时候，会面临有两种缺失值：

  • 补位的时候造成的空缺
  • 数据原先就存在缺失值
R 
df %>% complete(x, y)
  • 补位的时候造成的空缺，可通过fill = list(z = 0) 控制填充
R 
df %>% complete(x, y, fill = list(z = 0))
  • 数据原先就存在缺失值，最好通过 explicit = FALSE显式地控制是否填充
R 
df %>% complete(x, y, fill = list(z = 0), explicit = FALSE)

expand_grid() 与 crossing()

产生一个新的数据框，每行对应着向量元素的所有交叉组合

R 
expand_grid(x = 1:3, y = 1:2)
R 
crossing(x = 1:3, y = 1:2)

向量换成数据框也可以，其结果就是数据框行与元素的交叉组合

R 
expand_grid(df = data.frame(x = 1:2, y = c(2, 1)), z = 1:3)
R 
crossing(df = data.frame(x = 1:2, y = c(2, 1)), z = 1:3)

crossing()可以看作是expand_grid() + distinct()， 即crossing()在完成交叉组合之后会自动去重，比如

R 
expand_grid(x = c(1, 1), y = c(1:2))  # 不考虑去重
R 
crossing(x = c(1, 1), y = c(1:2))    # 考虑去重

separate() 与 unite()

R 
tb <- tibble::tribble(
  ~day, ~price,
  1,   "30-45",
  2,   "40-95",
  3,   "89-65",
  4,   "45-63",
  5,   "52-42"
)
R 
tb1 <- tb %>%
  separate(price, into = c("low", "high"), sep = "-")
tb1
R 
tb1 %>%
  unite(col = "price", c(low, high), sep = ":", remove = FALSE)

有时候分隔符搞不定的，可以用正则表达式，将捕获的每组弄成一列

R 
dfc <- tibble(x = c("1-12week", "1-10wk", "5-12w", "01-05weeks"))
dfc
R 
dfc %>% tidyr::extract(
  x,
  c("start", "end", "letter"), "(\\d+)-(\\d+)([a-z]+)",
  remove = FALSE
)

删除缺失值所在行drop_na()与replace_na()

R 
df <- tibble::tribble(
    ~name,     ~type, ~score, ~extra,
  "Alice", "english",     80,     10,
  "Alice",    "math",     NA,      5,
    "Bob", "english",     NA,      9,
    "Bob",    "math",     69,     NA,
  "Carol", "english",     80,     10,
  "Carol",    "math",     90,      5
  )

df

如果score列中有缺失值NA，就删除所在的row

R 
df %>%
  filter(!is.na(score))

或者用across()

R 
df %>%
  filter(
    across(score, ~ !is.na(.x))
  )

所有列，如果有缺失值NA，就删除所在的row

R 
df %>%
  filter(
    across(everything(), ~ !is.na(.x))
  )

现在有更简便的方法

R 
df %>%
  drop_na()

也可指定某一列

R 
df %>%
  drop_na(score)

没来参加考试，视为0分，可以用replace_na()

R 
df %>% mutate(score = replace_na(score, 0))

或者使用coalesce()

R 
df %>% mutate(score = coalesce(score, 0))
R 
df %>%
  mutate(
    across(c(score, extra), ~ coalesce(.x, 0))
  )

没来参加考试，用平均分代替

R 
df %>%
  mutate(
    score = replace_na(score, mean(score, na.rm = TRUE))
  )

当然也可以用if_else()来做

R 
df %>%
  mutate(
    score = if_else(is.na(score), mean(score, na.rm = TRUE), score)
  )
R 
pacman::p_unload(pacman::p_loaded(), character.only = TRUE)