如何透视表以从R中的变量行值创建列

时间:2011-06-24 17:03:21

标签: r pivot reshape

我有一个包含以下列的data.frame:Month,Store和Demand。

Month   Store   Demand
Jan     A   100
Feb     A   150
Mar     A   120
Jan     B   200
Feb     B   230
Mar     B   320

我需要转动它来制作一个新的data.frame或数组,每个月都有列,例如:

Store   Jan Feb Mar
A       100 150 120
B       200 230 320

非常感谢任何帮助。我刚开始用R。

3 个答案:

答案 0 :(得分:9)

> df <- read.table(textConnection("Month   Store   Demand
+ Jan     A   100
+ Feb     A   150
+ Mar     A   120
+ Jan     B   200
+ Feb     B   230
+ Mar     B   320"), header=TRUE)

因此,您的月份列很可能是按字母顺序排序的因子列 (编辑:)

> df$Month <- factor(df$Month, levels= month.abb[1:3])
 # Just changing levels was not correct way to handle the problem. 
 # Need to use within a factor(...) call.
> xtabs(Demand ~ Store+Month, df)
      Month
 Store Jan Feb Mar
     A 100 150 120
     B 200 230 320

一种稍微不那么明显的方法(因为'I'函数返回其参数):

> with(df, tapply(Demand, list(Store, Month) , I)  )
  Jan Feb Mar
A 100 150 120
B 200 230 320

答案 1 :(得分:5)

欢迎来到R.

通常有很多方法可以使用R来达到同一目的。另一种方法是使用Hadley的重塑包。

# create the data as explained by @Dwin
df <- read.table(textConnection("Month   Store   Demand
                                Jan     A   100
                                Feb     A   150
                                Mar     A   120
                                Jan     B   200
                                Feb     B   230
                                Mar     B   320"), 
                 header=TRUE)

# load the reshape package from Hadley -- he has created GREAT packages
library(reshape)

# reshape the data from long to wide
cast(df, Store ~ Month)

作为参考,你应该看看这个很棒的教程。 http://www.jstatsoft.org/v21/i12/paper

答案 2 :(得分:4)

如果数据在dat(并且级别设置为日历顺序),则另一个基本R解决方案是使用(令人难以置信的不直观)reshape()函数:

reshape(dat, v.names = "Demand", idvar = "Store", timevar = "Month", 
        direction = "wide")

对于数据片段给出:

> reshape(dat, v.names = "Demand", idvar = "Store", timevar = "Month", 
+         direction = "wide")
  Store Demand.Jan Demand.Feb Demand.Mar
1     A        100        150        120
4     B        200        230        320

如果您愿意,可以轻松清理名称:

> out <- reshape(dat, v.names = "Demand", idvar = "Store", timevar = "Month", 
+                direction = "wide")
> names(out)[-1] <- month.abb[1:3]
> out
  Store Jan Feb Mar
1     A 100 150 120
4     B 200 230 320

(为了得到上面的输出,我以与@ DWin的答案中显示的方式类似的方式读取数据,然后执行以下操作:

dat <- transform(dat, Month = factor(Month, levels = month.abb[1:3]))

其中dat就是我所说的数据)