如何从长到宽重塑和归纳分类数据？

Question

My data base is like this one:

db <- data.frame(var1 = c("A", "B", "C", "D", "E"), var2 = c("X", "X", "Y", "Y", "Y"),
           var3 = c("G", "H", "G", "G", "K"))
db

  var1 var2 var3
    A    X    G
    B    X    H
    C    Y    G
    D    Y    G
    E    Y    K

I'd like to reshape based in var2 and count the occurrences of var3 to get this result:

  var2 var3.G var3.H var3.K
    X    1      1      0
    Y    2      0      1

I have tried the cast and the reshape functions with no success.

Answer 1

The xtabs function is reasonably simple to use. The only cognitive jump is to realize that there is no LHS unless you want to do summation of a third variable:

> xtabs( ~var2+var3, data=db)
    var3
var2 G H K
   X 1 1 0
   Y 2 0 1

You don't want to do as.data.frame on this since it will convert to long form but you can use as.data.frame.matrix on it, since an R-'table' inherits from the 'matrix' class.

Answer 2

tbl <- data.frame( var2 = db[,2], var3 = paste("var3", db[,3], sep = "."))
table(tbl)
    var3
var2 var3.G var3.H var3.K
   X      1      1      0
   Y      2      0      1

Answer 3

One more option. Using the super useful data.table package:

library(data.table)

db <- data.table(var1 = c("A", "B", "C", "D", "E"), var2 = c("X", "X", "Y", "Y", "Y"),
           var3 = c("G", "H", "G", "G", "K"))

dcast.data.table(db, var2 ~ var3, fun = length, value.var= 'var3')
   var2 G H K
1:    X 1 1 0
2:    Y 2 0 1

Answer 4

Here is another way go about it:

You can use a combination of t() and table().

db <- data.frame(var1 = c("A", "B", "C", "D", "E"), 
                 var2 = c("X", "X", "Y", "Y", "Y"),
                 var3 = c("G", "H", "G", "G", "K"))
db

t(table(db$var3,db$var2))