R nest/unnest of dataframe 导致不相同的对象

Question

我第一次在 R 中使用 nest/unnest 函数，我不明白结果。我嵌套并立即取消嵌套并比较之前/之后的数据帧。为什么数据帧不相同？

> library(tidyverse)  
> concentration_original <- readRDS("./Data/concentration.Rds")
> print(concentration_original, n=15)
# A tibble: 12 x 5
   SUBJID    WT  DOSE  TIME  CONC
    <dbl> <dbl> <dbl> <dbl> <dbl>
 1      1  79.6  4.02 0      0.74
 2      1  79.6  4.02 0.25   2.84
 3      1  79.6  4.02 0.570  6.57
 4      1  79.6  4.02 1.12  10.5 
 5      1  79.6  4.02 2.02   9.66
 6      1  79.6  4.02 3.82   8.58
 7      2  72.4  4.4  0      0   
 8      2  72.4  4.4  0.27   1.72
 9      2  72.4  4.4  0.52   7.91
10      2  72.4  4.4  1      8.31
11      2  72.4  4.4  1.92   8.33
12      2  72.4  4.4  3.5    6.85
> 
> concentration_nested <- concentration_original %>% nest(data = c(TIME, CONC))
> concentration_nested
# A tibble: 2 x 4
  SUBJID    WT  DOSE data            
   <dbl> <dbl> <dbl> <list>          
1      1  79.6  4.02 <tibble [6 × 2]>
2      2  72.4  4.4  <tibble [6 × 2]>
> 
> concentration_unnested <- unnest(concentration_nested, cols = c(data))
> print(concentration_unnested, n=15)
# A tibble: 12 x 5
   SUBJID    WT  DOSE  TIME  CONC
    <dbl> <dbl> <dbl> <dbl> <dbl>
 1      1  79.6  4.02 0      0.74
 2      1  79.6  4.02 0.25   2.84
 3      1  79.6  4.02 0.570  6.57
 4      1  79.6  4.02 1.12  10.5 
 5      1  79.6  4.02 2.02   9.66
 6      1  79.6  4.02 3.82   8.58
 7      2  72.4  4.4  0      0   
 8      2  72.4  4.4  0.27   1.72
 9      2  72.4  4.4  0.52   7.91
10      2  72.4  4.4  1      8.31
11      2  72.4  4.4  1.92   8.33
12      2  72.4  4.4  3.5    6.85
> 
> if (identical(concentration_unnested, concentration_original)) {
+   print("After nest/unnest, we have a dataframe which IS IDENTICAL to the original")
+ } else {
+   print("After nest/unnest, we have a dataframe which IS NOT IDENTICAL to the original")
+ }
[1] "After nest/unnest, we have a dataframe which IS NOT IDENTICAL to the original"
> 
> all.equal(concentration_unnested, concentration_original)
[1] "Attributes: < Length mismatch: comparison on first 2 components >"
> 

请注意，我正在使用 all.equal 以查看问题可能与属性有关。如果我改用 all_equal，结果为 TRUE，但我仍然坚持 identical 函数说数据帧不一样。感谢您对此的任何帮助!

添加了原始 df 和嵌套/未嵌套 df 的 dput。

> dput(concentration_original)
structure(list(SUBJID = c(1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2), 
    WT = c(79.6, 79.6, 79.6, 79.6, 79.6, 79.6, 72.4, 72.4, 72.4, 
    72.4, 72.4, 72.4), DOSE = c(4.02, 4.02, 4.02, 4.02, 4.02, 
    4.02, 4.4, 4.4, 4.4, 4.4, 4.4, 4.4), TIME = c(0, 0.25, 0.57, 
    1.12, 2.02, 3.82, 0, 0.27, 0.52, 1, 1.92, 3.5), CONC = c(0.74, 
    2.84, 6.57, 10.5, 9.66, 8.58, 0, 1.72, 7.91, 8.31, 8.33, 
    6.85)), spec = structure(list(cols = list(SUBJID = structure(list(), class = c("collector_double", 
"collector")), WT = structure(list(), class = c("collector_double", 
"collector")), DOSE = structure(list(), class = c("collector_double", 
"collector")), TIME = structure(list(), class = c("collector_double", 
"collector")), CONC = structure(list(), class = c("collector_double", 
"collector"))), default = structure(list(), class = c("collector_guess", 
"collector")), skip = 1), class = "col_spec"), row.names = c(NA, 
-12L), class = c("tbl_df", "tbl", "data.frame"))
> dput(concentration_unnested)
structure(list(SUBJID = c(1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2), 
    WT = c(79.6, 79.6, 79.6, 79.6, 79.6, 79.6, 72.4, 72.4, 72.4, 
    72.4, 72.4, 72.4), DOSE = c(4.02, 4.02, 4.02, 4.02, 4.02, 
    4.02, 4.4, 4.4, 4.4, 4.4, 4.4, 4.4), TIME = c(0, 0.25, 0.57, 
    1.12, 2.02, 3.82, 0, 0.27, 0.52, 1, 1.92, 3.5), CONC = c(0.74, 
    2.84, 6.57, 10.5, 9.66, 8.58, 0, 1.72, 7.91, 8.31, 8.33, 
    6.85)), row.names = c(NA, -12L), class = c("tbl_df", "tbl", 
"data.frame"))
> 

附加信息:我想我发现了问题。原始小标题上的 spec= 信息包含与使用 read_csv 创建小标题时相关的信息。当 tibble 通过 nest/unnest 转换时，spec= 信息已被丢弃。还有另一个线程提到 spec= info 与 tibble 的内容不同步:Remove attributes from data read in readr::read_csv .在这种情况下，他们建议删除 spec= 属性:

attr(df, 'spec') <- NULL

Answer 1

根据我能够找到的内容，您的原始数据框与输出不相同的原因是原始数据框属于 col_spec 类，而输出不是。

使用新的 waldo 包，tidyverse 的一部分，我运行了以下命令:

compare(df, df %>% nest(data = c(TIME, CONC)) %>% unnest(cols = c(data)))
`attr(old, 'spec')` is an S3 object of class <col_spec>
`attr(new, 'spec')` is absent

看起来您使用 readr 读取了数据，并且生成的 df 是类 col_spec 的对象。嵌套原始 df 会删除此属性。

attr(df %>% nest(data = c(TIME, CONC)), 'spec')
NULL

因此，当您unnest 时，df 不相同。