如何重新排列可在 dplyr 中輕松查看的匯總統計信息？

Question

我的數據框中有一個匯總統計信息：

  war_3 a1_1_area_mean a1_2_area_mean a1_3_area_mean a1_4_area_mean a1_5_area_mean a1_6_area_mean
1     1     0.23827851     0.07843460     0.02531607      0.1193928      0.7635068     0.02333938
2     2     0.23162416     0.05949285     0.01422585      0.3565457      0.8593997     0.06895526
3     3     0.09187454     0.07274503     0.10357251      0.2821142      0.5929178     0.02455053
  a1_7_area_mean a1_8_area_mean a1_t_area_mean a2_1_area_mean a2_2_area_mean a2_3_area_mean
1    0.005387169      0.2725867       1.526242    0.107725394     0.19406917     0.02213419
2    0.016701786      0.2222106       1.829156    0.073991405     0.03504120     0.00815826
3    0.028382414      0.1997225       1.395880    0.003634443     0.03508602     0.00000000
  a2_4_area_mean a2_5_area_mean a2_t_area_mean a1_1_area_var a1_2_area_var a1_3_area_var a1_4_area_var
1     0.02024704   0.0040841950     0.34826000     1.2730028    0.13048871    0.05165589     0.1851353
2     0.07621595   0.0005078053     0.19391462     0.6114136    0.09287735    0.05697542     0.7284144
3     0.00000000   0.0000000000     0.03872046     0.1171754    0.07581946    0.35349703     0.3883895
  a1_5_area_var a1_6_area_var a1_7_area_var a1_8_area_var a1_t_area_var a2_1_area_var a2_2_area_var
1     2.7640424    0.01688505   0.001459156     0.8844626      7.940393    0.57992528    1.41104857
2     2.6797714    0.05490461   0.003428341     0.5725653      8.190389    0.18087732    0.11406984
3     0.9938991    0.01801805   0.006360622     0.3405592      3.460435    0.00306776    0.06579978
  a2_3_area_var a2_4_area_var a2_5_area_var a2_t_area_var a1_1_area_sd a1_2_area_sd a1_3_area_sd
1   0.067049470    0.06260921  0.0045015472    2.10734089    1.1282743    0.3612322    0.2272793
2   0.009580693    0.29505206  0.0005616327    0.85060972    0.7819294    0.3047579    0.2386952
3   0.000000000    0.00000000  0.0000000000    0.06861217    0.3423089    0.2753533    0.5945562
  a1_4_area_sd a1_5_area_sd a1_6_area_sd a1_7_area_sd a1_8_area_sd a1_t_area_sd a2_1_area_sd
1    0.4302735    1.6625410    0.1299425   0.03819890    0.9404587     2.817870   0.76152825
2    0.8534719    1.6370007    0.2343173   0.05855204    0.7566805     2.861886   0.42529674
3    0.6232090    0.9969449    0.1342313   0.07975351    0.5835745     1.860224   0.05538736
  a2_2_area_sd a2_3_area_sd a2_4_area_sd a2_5_area_sd a2_t_area_sd
1    1.1878757   0.25893912    0.2502183   0.06709357    1.4516683
2    0.3377423   0.09788102    0.5431869   0.02369879    0.9222851
3    0.2565147   0.00000000    0.0000000   0.00000000    0.2619392

以上匯總表來自以下腳本和原始數據框，如下所示：

  uid war_3 a1_1_area a1_2_area a1_3_area a1_4_area a1_5_area a1_6_area a1_7_area a1_8_area a1_t_area
1 1001     1         0   0.00000         0   0.67048    0.0000   0.02088         0   0.00000   0.69136
2 1002     2         0   0.00000         0   0.00000    0.9019   0.14493         0   0.00000   1.04683
3 1003     2         0   0.00000         0   0.00000    0.9019   0.00000         0   0.00000   0.90190
4 1004     2         0   1.09322         0   0.00000    0.0000   0.00000         0   0.00000   1.09322
5 1005     3         0   1.75000         0   0.00000    0.0000   0.00000         0   0.00000   1.75000
6 1006     2         0   2.43442         0   0.32223    0.0000   0.00000         0   0.76801   3.52466
  a2_1_area a2_2_area a2_3_area a2_4_area a2_5_area a2_t_area
1         0         0         0         0         0         0
2         0         0         0         0         0         0
3         0         0         0         0         0         0
4         0         0         0         0         0         0
5         0         0         0         0         0         0
6         0         0         0         0         0         0

summary <- df.anov %>% select(-uid) %>% group_by(war_3,) %>%
    summarize_each(funs(min,max,mean,median,var,sd)))

但是，由於很難通過均值、var 和 sd 比較成對的war_3 （組）中的每個值，因此我想將其轉換為以下格式：

    variable  war_3  mean  variance  s.d.
    a1_1_area, 1   , x    , x      , x
    a1_1_area, 2   , x    , x      , x
    a1_1_area, 3   , x    , x      , x
    a1_2_area, 1   , x    , x      , x
    a1_2_area, 2   , x    , x      , x
    a1_2_area, 3   , x    , x      , x
    a1_3_area, 1   , x    , x      , x
    a1_3_area, 2   , x    , x      , x
    a1_3_area, 3   , x    , x      , x
    a1_4_area, 1   , x    , x      , x
    a1_4_area, 2   , x    , x      , x
    a1_4_area, 3   , x    , x      , x
(it continues until `a2_5_area` in `variable`)

我曾經在dplyr使用gather將寬格式重新排列為簡單數據幀的長格式，但是這個數據幀需要更復雜的操作，這可能需要重復select(matches())左右。

變量是：
war_3變量對每條記錄進行分組group_by(war_3) %>% summarize_each(funs(mean,var,sd))在之前的操作中已經通過group_by(war_3) %>% summarize_each(funs(mean,var,sd))分組）
aX_Y_area_Z ：其中X具有兩個值1和2， Y傳播1-8為X=1和1-5為X=2 。 Z具有三個統計量，分別mean, variance and sd 。

你能幫我讓它成為可能嗎？
我更喜歡使用dplyr管道而不是data.table()解決方案。

以下腳本是非常手動的方式，但在每個gather()中都會生成重復的記錄，我不想手動指定每個列號或名稱。

 summary %>%
  gather(key1,mean,
    a1_1_area_mean,a1_2_area_mean,a1_3_area_mean,a1_4_area_mean,
    a1_5_area_mean,a1_6_area_mean,a1_7_area_mean,a1_8_area_mean,
    a1_t_area_mean,a2_1_area_mean,a2_2_area_mean,a2_3_area_mean,
    a2_4_area_mean,a2_5_area_mean,a2_t_area_mean) %>%
  gather(key2,var,
    a1_1_area_var,a1_2_area_var,a1_3_area_var,a1_4_area_var,
    a1_5_area_var,a1_6_area_var,a1_7_area_var,a1_8_area_var,
    a1_t_area_var,a2_1_area_var,a2_2_area_var,a2_3_area_var,
    a2_4_area_var,a2_5_area_var,a2_t_area_var) %>%
  gather(key3,sd,
    a1_1_area_sd,a1_2_area_sd,a1_3_area_sd,a1_4_area_sd,
    a1_5_area_sd,a1_6_area_sd,a1_7_area_sd,a1_8_area_sd,
    a1_t_area_sd,a2_1_area_sd,a2_2_area_sd,a2_3_area_sd,
    a2_4_area_sd,a2_5_area_sd,a2_t_area_sd) %>%
    mutate_at(vars(key1),funs(str_sub(.,1,9))) %>% select(-key2,-key3) %>%
    rename(key=key1) -> summary2

Answer 1

由於您沒有提供易於復制和粘貼的示例數據，因此我自己制作了一些

library(tidyverse)
data <-  mtcars %>% 
  group_by(cyl) %>%
  mutate(disp_1 = disp, disp_2=disp, mpg_1 = mpg, mpg_2 = mpg, drat_1=drat, drat_2=drat) %>% 
  select(-disp, -mpg, -drat) %>% 
  summarise_at(vars(contains("mpg"),contains("disp"), contains("drat")), list(mean =mean, sd = sd))


data
# A tibble: 3 x 13
cyl mpg_1_mean mpg_2_mean disp_1_mean disp_2_mean drat_1_mean drat_2_mean mpg_1_sd
<dbl>      <dbl>      <dbl>       <dbl>       <dbl>       <dbl>       <dbl>    <dbl>
1     4       26.7       26.7        105.        105.        4.07        4.07     4.51
2     6       19.7       19.7        183.        183.        3.59        3.59     1.45
3     8       15.1       15.1        353.        353.        3.23        3.23     2.56
# ... with 5 more variables: mpg_2_sd <dbl>, disp_1_sd <dbl>, disp_2_sd <dbl>,
#   drat_1_sd <dbl>, drat_2_sd <dbl>

然后，簡單地收集、分離和傳播

data %>% 
  gather(key, value, -cyl) %>% 
  separate(key, into = letters[1:3]) %>% 
  spread(c, value)
# A tibble: 18 x 5
     cyl a     b       mean     sd
   <dbl> <chr> <chr>  <dbl>  <dbl>
 1     4 disp  1     105.   26.9  
 2     4 disp  2     105.   26.9  
 3     4 drat  1       4.07  0.365
 4     4 drat  2       4.07  0.365
 5     4 mpg   1      26.7   4.51 
 6     4 mpg   2      26.7   4.51 
 7     6 disp  1     183.   41.6  
 8     6 disp  2     183.   41.6  
 9     6 drat  1       3.59  0.476
10     6 drat  2       3.59  0.476
11     6 mpg   1      19.7   1.45 
12     6 mpg   2      19.7   1.45 
13     8 disp  1     353.   67.8  
14     8 disp  2     353.   67.8  
15     8 drat  1       3.23  0.372
16     8 drat  2       3.23  0.372
17     8 mpg   1      15.1   2.56 
18     8 mpg   2      15.1   2.56