将 tibble 扩展到完整网格并保留所有值答案

【问题标题】：Expanding tibble to full grid and keep all values将 tibble 扩展到完整网格并保留所有值
【发布时间】：2020-02-14 06:11:32
【问题描述】：

如果我有一个像下面这样的数据框，是否可以扩展并保留所有值（不覆盖源 tibble 或保存中间结果）？

library(tidyverse)

d <- tibble(a = c("a", "a", "a", "b", "b"),
            b = c("2019", "2020", "2021", "2019", "2020"),
            x = 1:5) %>%
  spread(key = b, value = x) # I did this on purpose because ..

# is it possible to simplify this?
d %>%
  gather(-a, key = "b", val = "x") %>% # .. otherwise I could just join
  expand(a, b) %>%
  left_join(d %>%
              gather(-a, key = "b", val = x),
            by = c("a", "b"))

# is it possible to do something like:
d %>%
  gather(-a, key = "b", val = "x") %>%
  expand(a, b, .keep_all = TRUE) # there is of course no .keep_all for expand

# want data set
# a     b         x
# <chr> <chr> <int>
#   1 a     2019      1
# 2 a     2020      2
# 3 a     2021      3
# 4 b     2019      4
# 5 b     2020      5
# 6 b     2021     NA

我可能遗漏了一些东西，但我找不到解决方案。

【问题讨论】：

标签： r dplyr

【解决方案1】：

仍然不确定，也许：

my_expand <- function(d, ...) {
  keep_vars = ensyms(...)
  expand(d, !!!keep_vars) %>%
    left_join(d, by = map_chr(keep_vars, expr_text))
}

d %>% 
  my_expand(a, b)

# a     b         x
# <chr> <chr> <int>
#   1 a     2019      1
# 2 a     2020      2
# 3 a     2021      3
# 4 b     2019      4
# 5 b     2020      5
# 6 b     2021     NA

【讨论】：

【解决方案2】：

这个怎么样？

d %>%
  spread(b, x) %>%
  gather(b, x, -a) %>%
  arrange(x)
# A tibble: 6 x 3
#  a     b         x
#  <chr> <chr> <int>
#1 a     2019      1
#2 a     2020      2
#3 a     2021      3
#4 b     2019      4
#5 b     2020      5
#6 b     2021     NA

使用pivot_wider 和pivot_longer：

d %>% pivot_wider(names_from = b, values_from = x) %>%
  pivot_longer(-a, names_to = "b", values_to = "x")

# # A tibble: 6 x 3
#   a     b         x
# <chr> <chr> <int>
# 1 a     2019      1
# 2 a     2020      2
# 3 a     2021      3
# 4 b     2019      4
# 5 b     2020      5
# 6 b     2021     NA

【讨论】：