在 R 中重新格式化数据框

Question

我有多个列，想重新格式化数据框以减少列数。

这是我的当前 df:


# Dataframe
df <- data.frame(
    ~Location, ~Product_Name, ~Category, ~Machine1, ~Machine2 ~Machine1_adds, ~Machine2_adds, ~Sales1, ~Saless2, Spoils1, Spoils2
    A,  "Snickers",   Candy,  0, 1,  .5, , 2, 1
    A,  "Kitcat",   Candy,  0, 1,  , , 2, 1
    A,  "Pepsi",   Bev,  1, 1,  , , 3, 0
    B,  "Coke",   Bev,  1, 0, ,  .45, 1, 1
    B,  "Gatoraid",   Bev,  0, 1, ,  .45, 1, 0
    B,  "Sprite",   Bev,  1, 1,  , , 1, 0
)
df

我想将数据框重新格式化为 machine 是一个列，sales 是另一个列，而 spoils 是最后一个，每个值行都调整为新列。参考输出：

# Dataframe
new_df <- data.frame(
    ~Location, ~Product_Name, ~Category, ~Machine, ~Machine_adds, ~Sales, Spoils
    A,  "Snickers",   Candy, 1,  0,  .5, 2
    A,  "Kitcat",   Candy, 1,  0,  , 2
    A,  "Pepsi",   Bev,  1, 1,  , 3
    B,  "Coke",   Bev,  2, 1, .45, 1
    B,  "Gatoraid",   Bev, 2,  0,  .45, 1
    B,  "Sprite",   Bev,  2, 1,  , 0
)
new_df


I implemented the melt function but I am not getting the 'Machine' Column to represent which machine I am pulling from (either Machine 1, 2, 3 etc..).

Answer 1

这是 data.table

中带有 melt 的选项

library(data.table)
melt(setDT(df), measure = patterns("^Machine", "^Sales", "^Spoils"), 
      value.name = c("Machine_adds", "Sales", "Spoils"))[, variable := NULL][]
#    Location Product_Name Category Machine_adds Sales Spoils
# 1:        A     Snickers    Candy       0  .5      2
# 2:        A       Kitcat    Candy       0          2
# 3:        A        Pepsi      Bev       1          3
# 4:        B         Coke      Bev       1          1
# 5:        B     Gatoraid      Bev       0          1
# 6:        B       Sprite      Bev       1          1
# 7:        A     Snickers    Candy       1          1
# 8:        A       Kitcat    Candy       1          1
# 9:        A        Pepsi      Bev       1          0
#10:        B         Coke      Bev       0 .45      1
#11:        B     Gatoraid      Bev       1 .45      0
#12:        B       Sprite      Bev       1          0

更新

根据 OP 的更新示例，如果有 'Machine' 和 Machine_adds' 列，我们可以将 patterns 稍微更改为

# creating new columns in the dataset
df[c('Machine1', 'Machine2')] <- df[c("Machine1_adds", "Machine2_adds")]
melt(setDT(df), measure = patterns("^Machine\d+$", 
         "^Machine\d+_adds$", "^Sales", "^Spoils"), 
   value.name = c("Machine", "Machine_adds", "Sales", "Spoils"))[, 
           variable := NULL][]

或使用 tidyr

中的 pivot_longer

library(dplyr)
library(tidyr)
library(stringr)
df %>% 
     rename_at(3:ncol(.), ~
           str_replace(., "(\d+)_?.*", "_\1")) %>% 
     pivot_longer(cols =  matches("^(Machine|Sales|Spoils)"),
        names_to = c(".value", "group"), names_sep = "_") %>%
     select(-group)
# A tibble: 12 x 6
#   Location Product_Name Category Machine Sales Spoils
#   <chr>    <chr>        <chr>      <dbl> <chr>  <dbl>
# 1 A        Snickers     Candy          0 .5       2
# 2 A        Snickers     Candy          1          1
# 3 A        Kitcat       Candy          0          2
# 4 A        Kitcat       Candy          1          1
# 5 A        Pepsi        Bev            1          3
# 6 A        Pepsi        Bev            1          0
# 7 B        Coke         Bev            1          1
# 8 B        Coke         Bev            0 .45      1
# 9 B        Gatoraid     Bev            0          1
#10 B        Gatoraid     Bev            1 .45      0
#11 B        Sprite       Bev            1          1
#12 B        Sprite       Bev            1          0

更新

df %>%
   rename_at(vars(matches('^Machine.*adds$')), ~ 
         str_replace(., '(\d+)_(\w+)$', '_\2\1')) %>% 
   rename_at(3:ncol(.), ~ str_replace(., "(\d+)_?.*", ":\1")) %>%
   pivot_longer(cols =  matches("^(Machine|Sales|Spoils)"),
        names_to = c(".value", "group"), names_sep = ":") %>%
   select(-group)

数据

df <- structure(list(Location = c("A", "A", "A", "B", "B", "B"), 
     Product_Name = c("Snickers", 
"Kitcat", "Pepsi", "Coke", "Gatoraid", "Sprite"), Category = c("Candy", 
"Candy", "Bev", "Bev", "Bev", "Bev"), Machine1_adds = c(0, 0, 
1, 1, 0, 1), Machine2_adds = c(1, 1, 1, 0, 1, 1), Sales1 = c(".5", 
"", "", "", "", ""), Sales2 = c("", "", "", ".45", 
".45", ""), Spoils1 = c(2, 2, 3, 1, 1, 1), Spoils2 = c(1, 
1, 0, 1, 0, 0)), row.names = c(NA, -6L), class = c("tbl_df", 
"tbl", "data.frame"))

Answer 2

根据您所拥有的，您正在寻找变量对齐为 list(c(4,5),c(6,7),c(8,9)) 的 reshape 函数。您可以使用：

reshape(df,t(matrix(4:ncol(df),2)),idvar = 1:3,dir="long")

或

reshape(df,list(c(4,5),c(6,7),c(8,9)),idvar = 1:3,dir="long")

为了得到你的名字，我将使用 v.names 参数

reshape(df,list(c(4,5),c(6,7),c(8,9)),idvar = 1:3,dir="long",
         v.names = c("Machine_adds","Sales","Spoils"))[-4]# -4 removes the time variable.
                   Location Product_Name Category Machine_adds Sales Spoils
A.Snickers.Candy.1        A     Snickers    Candy            0  .5      2
A.Kitcat.Candy.1          A       Kitcat    Candy            0          2
A.Pepsi.Bev.1             A        Pepsi      Bev            1          3
B.Coke.Bev.1              B         Coke      Bev            1          1
B.Gatoraid.Bev.1          B     Gatoraid      Bev            0          1
B.Sprite.Bev.1            B       Sprite      Bev            1          1
A.Snickers.Candy.2        A     Snickers    Candy            1          1
A.Kitcat.Candy.2          A       Kitcat    Candy            1          1
A.Pepsi.Bev.2             A        Pepsi      Bev            1          0
B.Coke.Bev.2              B         Coke      Bev            0 .45      1
B.Gatoraid.Bev.2          B     Gatoraid      Bev            1 .45      0
B.Sprite.Bev.2            B       Sprite      Bev            1          0

在 R 中重新格式化数据框

Reformat dataframe in R

r

dataframe

melt

dplyr

更新

更新

数据