tidyr 完整案例嵌套误区

Question

我不理解 tidyr 的 nesting + complete 的使用会导致错误的结果。我已阅读帮助手册并尝试了示例，但仍然无法在 tidyverse 中生成我想要的内容。

这是我拥有的数据：

##   project_id meta  domain1 question_count tag_count
##        <int> <chr> <chr>            <int>     <int>
## 1          1 A     d                    3         2
## 2          1 A     e                    3         1
## 3          1 B     h                    3         3
## 4          1 B     i                    3         2
## 5          2 A     d                    2         1
## 6          2 B     i                    2         1
## 7          2 C     k                    2         2

期望的输出

##    project_id meta  domain1 question_count tag_count
##         <int> <chr> <chr>            <int>     <int>
##  1          1 A     d                    3         2
##  2          1 A     e                    3         1
##  3          1 B     h                    3         3
##  4          1 B     i                    3         2
##  5          1 C     k                    3         0
##  6          2 A     d                    2         1
##  7          2 A     e                    2         0
##  8          2 B     h                    2         1
##  9          2 B     i                    2         1
## 10          2 C     k                    2         2

MWE 和我的失败尝试

## MWE
library(dplyr); library(tidyr)

dat <- data_frame(
    project_id = as.integer(c(rep(1, 4), rep(2, 3))),
    meta = c('A', 'A', 'B', 'B', 'A', 'B', 'C'),
    domain1 = c('d', 'e', 'h', 'i', 'd', 'i', 'k'),
    question_count =  as.integer(c(rep(3, 4), rep(2, 3))),
    tag_count =  as.integer(c(2, 1, 3, 2, 1, 1, 2))
)


## Failed Attempts
dat %>%
    complete(
        project_id, 
        tidyr::nesting(question_count, meta, domain1), 
        fill = list(tag_count = 0)
    )

## # A tibble: 14 x 5
##    project_id question_count meta  domain1 tag_count
##         <int>          <int> <chr> <chr>       <dbl>
##  1          1              2 A     d            0   
##  2          1              2 B     i            0   
##  3          1              2 C     k            0   
##  4          1              3 A     d            2.00
##  5          1              3 A     e            1.00
##  6          1              3 B     h            3.00
##  7          1              3 B     i            2.00
##  8          2              2 A     d            1.00
##  9          2              2 B     i            1.00
## 10          2              2 C     k            2.00
## 11          2              3 A     d            0   
## 12          2              3 A     e            0   
## 13          2              3 B     h            0   
## 14          2              3 B     i            0   

dat %>%
    complete(
        project_id, question_count,
        tidyr::nesting(meta, domain1), 
        fill = list(tag_count = 0)
    )

## # A tibble: 20 x 5
##    project_id question_count meta  domain1 tag_count
##         <int>          <int> <chr> <chr>       <dbl>
##  1          1              2 A     d            0   
##  2          1              2 A     e            0   
##  3          1              2 B     h            0   
##  4          1              2 B     i            0   
##  5          1              2 C     k            0   
##  6          1              3 A     d            2.00
##  7          1              3 A     e            1.00
##  8          1              3 B     h            3.00
##  9          1              3 B     i            2.00
## 10          1              3 C     k            0   
## 11          2              2 A     d            1.00
## 12          2              2 A     e            0   
## 13          2              2 B     h            0   
## 14          2              2 B     i            1.00
## 15          2              2 C     k            2.00
## 16          2              3 A     d            0   
## 17          2              3 A     e            0   
## 18          2              3 B     h            0   
## 19          2              3 B     i            0   
## 20          2              3 C     k            0   

dat %>%
    complete(
        meta, domain1,
        tidyr::nesting(project_id, question_count), 
        fill = list(tag_count = 0)
    )

## # A tibble: 30 x 5
##    meta  domain1 project_id question_count tag_count
##    <chr> <chr>        <int>          <int>     <dbl>
##  1 A     d                1              3      2.00
##  2 A     d                2              2      1.00
##  3 A     e                1              3      1.00
##  4 A     e                2              2      0   
##  5 A     h                1              3      0   
##  6 A     h                2              2      0   
##  7 A     i                1              3      0   
##  8 A     i                2              2      0   
##  9 A     k                1              3      0   
## 10 A     k                2              2      0   
## # ... with 20 more rows

Answer 1

我们可以为两组列使用两个nesting。

library(dplyr)
library(tidyr)

dat %>%
  complete(
    nesting(meta, domain1),
    nesting(project_id, question_count),
    fill = list(tag_count = 0)
  ) %>%
  arrange(project_id, meta) %>%
  select(names(dat))
# # A tibble: 10 x 5
#    project_id meta  domain1 question_count tag_count
#         <int> <chr> <chr>            <int>     <dbl>
#  1          1 A     d                    3      2.00
#  2          1 A     e                    3      1.00
#  3          1 B     h                    3      3.00
#  4          1 B     i                    3      2.00
#  5          1 C     k                    3      0   
#  6          2 A     d                    2      1.00
#  7          2 A     e                    2      0   
#  8          2 B     h                    2      0   
#  9          2 B     i                    2      1.00
# 10          2 C     k                    2      2.00