如何重塑数据集（从长到宽），每个类别有两个测量列，而无需在 R 中进行额外计算

我有长格式的数据集（请参阅下面的代码来生成此数据集）：

region   week   average   percent
    A      20         5        30
    A      21         7        40
    A      22        15        50
    B      20         4        15
    B      21         8        27
    B      22         3        11
    ...

我想准备它进行演示（带有 RMarkdown 的 HTML），所以我需要将其转换为宽的average and percent对于每个week像这样：

                 20                 21                  22
region    average  percent   average  percent    average  percent
     A          5       30        7        40         15       50
     B          4       15        8        27          3       11

我探索过dcast, dyplr, tidyr, htmlTable，还有许多其他人没有成功。我不需要执行任何计算，只需以不同的格式组织数据集即可。

我过去曾这样做过，但我必须编写大量代码。我想知道在 R 中是否有一种简单的方法可以做到这一点。

您可以使用此代码来创建测试数据集：

region = c( "A", "A", "A", "B", "B", "B" )
week = c( "20", "21", "22", "20", "21", "22" )
average = c( 5, 7, 15, 4, 8, 3 )
percent = c( 30, 40, 50, 15, 27, 11 )

test = data.frame(
  region,
  week,
  average,
  percent
)

任何帮助表示赞赏。

谢谢。

@akrun 和 @Matt L. 的两个答案都很好地解决了问题。 @Matt L. 绝对是最简单的一个。谢谢。我在这里发布我根据 @akrun 的答案提出的解决方案。我肯定会在我的最终代码中使用 tidyr 。

library(htmlTable)
library(data.table)
library(Hmisc)

# Create input (initial) dataset (long)
region = c( "A", "A", "A", "B", "B", "B" )
week = c( "20", "21", "22", "20", "21", "22" )
average = c( 5, 7, 15, 4, 8, 3 )
percent = c( 30, 40, 50, 15, 27, 11 )

input_ds = data.frame(
  region,
  week,
  average,
  percent
)

# Reshape the dataset into wide, using columns average and percent
# for each week
reshaped_ds = dcast(
  setDT( input_ds ),
  region ~ week,
  value.var = c("average", "percent")
)

# Extract the week number from each column and get a list of indices 
# sorted by week number
col_order <- order(
  as.numeric(
    sub( ".*_", "", names( reshaped_ds )[-1] )
  )
)

# Re-order columns according to col_order
setcolorder(
  reshaped_ds, names(reshaped_ds)[c(1, col_order + 1)]
)

# Prepare the names for group columns
col_group_names = unique(
  paste(
    "Week",
    sub( ".*_", "", names(reshaped_ds)[-1] )
  )
)

# Create another dataset so we don't mess up the reshaped_ds
final_table_ds = reshaped_ds

# Remove '_##' from column names
names( final_table_ds ) = sub(
  "_.*", "", names( final_table_ds )
)

# Capitalize the first letter of each column name
names( final_table_ds ) = capitalize( names( final_table_ds ) )

# Generate the final table in HTML
htmlTable(
  final_table_ds,
  rnames = FALSE,
  cgroup = c( "", col_group_names ),
  n.cgroup = c( 1, rep( 2, length(col_group_names) ) ),
  col.rgroup = c( "none", "#EDEDED" )
)

最终输出：