计算测量的汇总统计数据并将它们转至 R [重复] 中的列

Question

我有一个这样的数据框

Step <- c("1","1","4","3","2","2","3","4","4","3","1","3","2","4","3","1","2")
Length <- c(0.1,0.5,0.7,0.8,0.2,0.1,0.3,0.8,0.9,0.15,0.25,0.27,0.28,0.61,0.15,0.37,0.18)
Breadth <- c(0.13,0.35,0.87,0.38,0.52,0.71,0.43,0.8,0.9,0.15,0.45,0.7,0.8,0.11,0.11,0.47,0.28)
Height <- c(0.31,0.35,0.37,0.38,0.32,0.51,0.53,0.48,0.9,0.15,0.35,0.32,0.22,0.11,0.17,0.27,0.38)
Width <- c(0.21,0.25,0.27,0.8,0.2,0.21,0.3,0.28,0.29,0.65,0.55,0.37,0.26,0.31,0.5,0.7,0.8)

df <- data.frame(Step,Length,Breadth,Height,Width) 

我正在尝试计算按步骤分组的测量值的 最大值、最小值、平均值、中值、标准偏差，然后将这些具有测量值的列旋转为一列。

我的想要的输出是

  Measurement max_1 min_1 mean_1 median_1       sd_1 max_2 min_2 mean_2 median_2       sd_2 max_3 min_3 mean_3 median_3      sd_3 max_4 min_4 mean_4 median_4       sd_4
       Length  0.50  0.10 0.3050     0.31 0.17058722  0.28  0.10 0.1900    0.190 0.07393691  0.80  0.15  0.334     0.27 0.2693139  0.90  0.61 0.7525    0.750 0.12526638
      Breadth  0.47  0.13 0.3500     0.40 0.15577760  0.80  0.28 0.5775    0.615 0.23012680  0.70  0.11  0.354     0.38 0.2383904  0.90  0.11 0.6700    0.835 0.37567720
       Height  0.35  0.27 0.3200     0.33 0.03829708  0.51  0.22 0.3575    0.350 0.12120919  0.53  0.15  0.310     0.32 0.1570032  0.90  0.11 0.4650    0.425 0.32888701
        Width  0.70  0.21 0.4275     0.40 0.23669601  0.80  0.20 0.3675    0.235 0.28952547  0.80  0.30  0.524     0.50 0.2040343  0.31  0.27 0.2875    0.285 0.01707825

我正在尝试以这种方式计算汇总统计信息，但这不是一种有效的方式。

library(dplyr)
df1 <- df %>%
  group_by(Step) %>%
  summarise(Length_Mean = mean(Length),
            Breadth_Mean = mean(Breadth),
            Height_Mean = mean(Height),
            Width_Mean = mean(Width))

如何以最少的代码高效地完成我想要的输出？有人能指出我正确的方向吗？

Answer 1

您可以使用summarize 的"scoped" 版本来计算相同的摘要 一次统计多列。来自?scoped：

以 _if、_at 或 _all 为后缀的变体应用一个表达式 （有时是几个）到指定子集中的所有变量。这 子集可以包含所有变量（_all 变体），一个 vars() 选择 （_at 变体），或使用谓词选择的变量（_if 变体）。

这里summarize_all可能是个不错的选择；它选择所有列，除了 用于分组列。您还可以提供几个汇总函数 计算选择中的每个变量。

library(tidyverse)

# Calculate the summary statistics
sums <- df %>% 
  group_by(Step) %>% 
  summarize_all(funs(max, min, mean, median, sd))

sums
#> # A tibble: 4 x 21
#>   Step  Length_max Breadth_max Height_max Width_max Length_min Breadth_min
#>   <fct>      <dbl>       <dbl>      <dbl>     <dbl>      <dbl>       <dbl>
#> 1 1           0.5         0.47       0.35      0.7        0.1         0.13
#> 2 2           0.28        0.8        0.51      0.8        0.1         0.28
#> 3 3           0.8         0.7        0.53      0.8        0.15        0.11
#> 4 4           0.9         0.9        0.9       0.31       0.61        0.11
#> # ... with 14 more variables: Height_min <dbl>, Width_min <dbl>,
#> #   Length_mean <dbl>, Breadth_mean <dbl>, Height_mean <dbl>,
#> #   Width_mean <dbl>, Length_median <dbl>, Breadth_median <dbl>,
#> #   Height_median <dbl>, Width_median <dbl>, Length_sd <dbl>,
#> #   Breadth_sd <dbl>, Height_sd <dbl>, Width_sd <dbl>

现在我们有了汇总统计数据，剩下要做的就是 重塑数据以实现所需的输出。为此，gather、spread、 tidyr 的separate 和unite 派上用场了：

sums %>% 
  # Reshape to long format
  gather(col, val, -Step) %>% 
  # Separate the measurement and the summary statistic
  separate(col, into = c("Measurement", "stat")) %>% 
  arrange(Step) %>% 
  # Create the desired column headings
  unite(col, stat, Step) %>% 
  # Need to use factors to preserve order
  mutate_at(vars(col, Measurement), fct_inorder) %>% 
  # Reshape back to wide format
  spread(col, val)
#> # A tibble: 4 x 21
#>   Measurement max_1 min_1 mean_1 median_1   sd_1 max_2 min_2 mean_2
#>   <fct>       <dbl> <dbl>  <dbl>    <dbl>  <dbl> <dbl> <dbl>  <dbl>
#> 1 Length       0.5   0.1   0.305    0.31  0.171   0.28  0.1   0.19 
#> 2 Breadth      0.47  0.13  0.35     0.4   0.156   0.8   0.28  0.578
#> 3 Height       0.35  0.27  0.32     0.330 0.0383  0.51  0.22  0.358
#> 4 Width        0.7   0.21  0.428    0.4   0.237   0.8   0.2   0.368
#> # ... with 12 more variables: median_2 <dbl>, sd_2 <dbl>, max_3 <dbl>,
#> #   min_3 <dbl>, mean_3 <dbl>, median_3 <dbl>, sd_3 <dbl>, max_4 <dbl>,
#> #   min_4 <dbl>, mean_4 <dbl>, median_4 <dbl>, sd_4 <dbl>

由reprex package (v0.2.0) 于 2018 年 5 月 24 日创建。