使用 ftransform 和来自折叠 R 包的 fgroup_by

问题描述

我正在尝试使用 dplyr 包 R 重现 collapse 代码的以下输出。

dplyr 代码

library(tidyverse)
starwars %>%
  select(name,mass,species) %>%
  group_by(species) %>%
  mutate(mass_norm = mean(mass,na.rm = TRUE))

dplyr 代码输出

# A tibble: 87 x 4
# Groups:   species [38]
   name                mass species mass_norm
   <chr>              <dbl> <chr>       <dbl>
 1 Luke Skywalker        77 Human        82.8
 2 C-3PO                 75 Droid        69.8
 3 R2-D2                 32 Droid        69.8
 4 Darth Vader          136 Human        82.8
 5 Leia Organa           49 Human        82.8
 6 Owen lars            120 Human        82.8
 7 Beru Whitesun lars    75 Human        82.8
 8 R5-D4                 32 Droid        69.8
 9 Biggs Darklighter     84 Human        82.8
10 Obi-Wan Kenobi        77 Human        82.8
# … with 77 more rows

collapse 代码

library(collapse)
starwars %>%
  fselect(name,species) %>%
  fgroup_by(species) %>%
  ftransform(mass_norm = fmean(mass,na.rm = TRUE))

collapse 代码输出

# A tibble: 87 x 4
   name                mass species mass_norm
 * <chr>              <dbl> <chr>       <dbl>
 1 Luke Skywalker        77 Human        97.3
 2 C-3PO                 75 Droid        97.3
 3 R2-D2                 32 Droid        97.3
 4 Darth Vader          136 Human        97.3
 5 Leia Organa           49 Human        97.3
 6 Owen lars            120 Human        97.3
 7 Beru Whitesun lars    75 Human        97.3
 8 R5-D4                 32 Droid        97.3
 9 Biggs Darklighter     84 Human        97.3
10 Obi-Wan Kenobi        77 Human        97.3
# … with 77 more rows

Grouped by:  species  [38 | 2 (5.5)] 

想知道为什么我用 collapse 代码得到错误答案。任何提示。

解决方法

fmean 默认使用 na.rm = TRUE。此外，还有一个选项可以指定 fmean 内的分组，即 g。默认情况下，TRA 为 NULL 并返回汇总输出，但我们可以将其更改为 replace_fill 以返回完整长度

library(collapse)
ftransform(slt(starwars,name,mass,species),mass_norm = fmean(mass,species,TRA = 'replace_fill'))

-输出

# A tibble: 87 x 4
#   name                mass species mass_norm
# * <chr>              <dbl> <chr>       <dbl>
# 1 Luke Skywalker        77 Human        82.8
# 2 C-3PO                 75 Droid        69.8
# 3 R2-D2                 32 Droid        69.8
# 4 Darth Vader          136 Human        82.8
# 5 Leia Organa           49 Human        82.8
# 6 Owen Lars            120 Human        82.8
# 7 Beru Whitesun lars    75 Human        82.8
# 8 R5-D4                 32 Droid        69.8
# 9 Biggs Darklighter     84 Human        82.8
#10 Obi-Wan Kenobi        77 Human        82.8
# … with 77 more rows

如果我们要使用链式，使用GRP来指定数据上的g或分组变量（.）

library(dplyr)
starwars %>%
 fselect(name,species) %>%
 fgroup_by(species) %>%
 ftransform(mass_norm = fmean(mass,GRP(.),TRA = 'replace'))