如何通过多个变量（列）将长格式数据转换为宽格式数据并相互堆叠？

Question

我有20个站的月度时间序列数据（1987-2017）。我想将长格式数据转换为宽格式数据，以便覆盖 20 个站的所有数据都在一个数据帧中。

head(Monthly_rainfall2[-1:-20,1:5]) #long format data

# A tibble: 6 x 5
  Year  Month  stn1    stn2   stn3
  <chr> <ord> <dbl>    <dbl>  <dbl>
1 1987  Jan      NA       NA    0  
2 1987  Feb      NA       NA   60.5
3 1987  Mar      NA       NA   66  
4 1987  Apr      NA       NA   64  
5 1987  May      NA       NA  183. 
6 1987  Jun      NA       NA  216

请注意，月份列是有序因子。

 dput(Monthly_rainfall2[21:50,1:4])
 structure(list(Year = c("1987", "1987", "1987", "1987", "1987", 
"1987", "1987", "1987", "1987", "1987", "1987", "1987", "1988", 
"1988", "1988", "1988", "1988", "1988", "1988", "1988", "1988", 
"1988", "1988", "1988", "1989", "1989", "1989", "1989", "1989", 
"1989"), Month = structure(c(1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 
9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 
12L, 1L, 2L, 3L, 4L, 5L, 6L), .Label = c("Jan", "Feb", "Mar", 
"Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
), class = c("ordered", "factor")), stn1 = c(NA_real_, NA_real_, 
NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, 
NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, 
NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, 
NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_
), stn2 = c(NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, 
NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, 
NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, 
NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, NA_real_, 
NA_real_, NA_real_, NA_real_, NA_real_)), class = c("tbl_df", 
"tbl", "data.frame"), row.names = c(NA, -30L))

我试过下面的代码

library(tidyr)

wide_data <- spread(Monthly_rainfall2[1:3], Month, stn1 )

上面的代码提供了我想要的，但是我无法在一个数据帧中同时为所有站生成

我希望我的数据框看起来像这样，其中 stn2 在 stn1 之后立即开始，随后所有站点都遵循相同的模式

Year     Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
#stn1
1987     0.8 0.5 0.8 2   20   25  30  30  21  22 3    0
1988     1   1.2  1.8 2  20   22   25 21   15  12  10  9
...
2017     0.5  1  14  19   17   14  15  13  10  14  18  10
#stn2
1987     0.8 0.5 0.8 2   20   25  30  30  21  22 3    0
1988     1   1.2  1.8 2  20   22   25 21   15  12  10  9
...
2017     0.5  1  14  19   17   14  15  13  10  14  18  10
#stn3
1987     0.8 0.5 0.8 2   20   25  30  30  21  22 3    0
1988     1   1.2  1.8 2  20   22   25 21   15  12  10  9
...
2017     0.5  1  14  19   17   14  15  13  10  14  18  10

Answer 1

我们可以 gather 转换成 'long' 格式，然后再做一个 spread

library(tidyverse)
df1 %>% 
   gather(key, val, Bhur:Chamkhar) %>%
   spread(Month, val)

注意：OP 的 dput 只有 "Bhur/Chamkhar" 列的 NA

Answer 2

使用 data.table

的解决方案

DT <- as.data.table(df)

#filling all ne NAs for optical rasons
DT[, c("stn1", "stn2")  := .(sample(1:100, 30), sample(1:100, 30))]

dcast(DT, Month ~ Year, value.var = c("stn1", "stn2")) %>%
  melt(id.vars = 1, variable.name = "year") %>%
  dcast(year ~ Month, value.var = "value") -> DT2

        year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
1: stn1_1987  98   5  89  84  75  36   4  34  26  78  33  28
2: stn1_1988  67  74  40  63   9  19  79  61  66  93  47  62
3: stn1_1989  68  29   7  46  54  87  NA  NA  NA  NA  NA  NA
4: stn2_1987  31  61  74  89  46  54  70  80  84   6  96  32
5: stn2_1988  75  71  11  99  20   7  77  13  52  14   2  41
6: stn2_1989  83  22  97  43  59  15  NA  NA  NA  NA  NA  NA

如果你想year和stn分开你可以这样做：

DT2[, c("stn", "year") := tstrsplit(year, "_", fixed=TRUE)]

   year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec  stn
1: 1987  98   5  89  84  75  36   4  34  26  78  33  28 stn1
2: 1988  67  74  40  63   9  19  79  61  66  93  47  62 stn1
3: 1989  68  29   7  46  54  87  NA  NA  NA  NA  NA  NA stn1
4: 1987  31  61  74  89  46  54  70  80  84   6  96  32 stn2
5: 1988  75  71  11  99  20   7  77  13  52  14   2  41 stn2
6: 1989  83  22  97  43  59  15  NA  NA  NA  NA  NA  NA stn2

如何通过多个变量（列）将长格式数据转换为宽格式数据并相互堆叠？

How to convert long format to wide format data over multiple variable (Column) and be stacked onto each other?

r

reshape2

tidyr