table-operations: Compute table margin or table slice

Description

For a contingency table in array form, compute the sum of table entries for a given index (i.e. a marginal table) or find the slice of the table defined by specific margins being at a specific level.

Usage

tableOp(t1, t2, op = "*")
tableMult(t1, t2)
tableDiv(t1, t2)
tableMargin(x, margin, keep.class=FALSE)
tableSlice(x, margin, level, impose)
tablePerm(x, perm, resize = TRUE, keep.class=FALSE)
tableSetSliceValue(x, margin, level, complement=FALSE, value=0)

Arguments

x, t1, t2

An array (a vector with a dim and a dimnames attribute)

margin

An index, either numerical or character

keep.class

If TRUE the result will be forced to have the same class as the input; otherwise the result will be an array.

level

A value, either numerical or character

impose

Possible value used to fill up a slice to give it full dimension

complement

If TRUE then the entries of the complement of the slice is set to some value.

value

The value to put into the entries given by margin and level - or their complement if complement=TRUE

Either "*" or "/"

perm

The subscript permutation vector, which must be a permutation of the integers 1:n, where n is the number of dimensions of a OR a permutation of the dimension names of a. The default is to reverse the order of the dimensions. A permutation of the dimensions of a.

resize

A flag indicating whether the vector should be resized as well as having its elements reordered.

Value

An array.

Details

tableMargin: tableMargin is analogous to margin.table except that margin can be given both as array indices or as variable names

tableSlice: If the table x has dimensions Z,U,V where V has levels 1 and 2 then tableSlice can extract the slice of x (in this case a 2-way table) defined by e.g. U=2. Setting impose=1000 implies that a 3-way table is returned with the U=2 slice in the right place and the U=1-slice consisting of 1000 in each cell.

tableOp: If t1 has dimnames A and B and t2 has dimnames B and C then tableOp(t1,t2) will return a table (an array) with dimnames A, B and C containing the product.

tableMult(t1,t2) is a wrapper for tableOp(t1,t2, op="*") and tableMult(t1,t2) is a wrapper for tableDiv(t1,t2, op="/")

tablePerm: A wrapper for aperm, but tablePerm accepts dimnames in addition to indices.

See examples below.

Examples

Run this code


data(HairEyeColor)

tableMargin(HairEyeColor, "Hair")           ## same as:
tableMargin(HairEyeColor, 1)
tableMargin(HairEyeColor, c("Hair", "Eye")) ## same as:
tableMargin(HairEyeColor, c(1, 2))


tableSlice(HairEyeColor, "Sex", "Male")              ## same as:
tableSlice(HairEyeColor, 3, 1)
tableSlice(HairEyeColor, "Sex", "Male", impose=1000) ## same as:
tableSlice(HairEyeColor, 3, 1, impose=1000)

t1 <- array(1:4, dim=c(2,2), dimnames=list(gender=c('male','female'),
                                           answer=c('yes','no')))
t2 <- array(5:8, dim=c(2,2), dimnames=list(answer=c('yes','no'),
                                           category=c(1,2)))

tableOp(t1, t2, "*")
tableOp(t1, t2, "/")

data(reinis)

t1 <- tableMargin(reinis, c(6, 5, 2, 1))
t2 <- tableMargin(reinis, c(6, 5, 3, 4))
tt1 <- tableOp(t1, t2)

t1 <- tableMargin(reinis, c(6, 5, 2, 4, 1))
t2 <- tableMargin(reinis, c(6, 5, 4))
tt1 <- tableOp2(t1, t2)

tableSetSliceValue(HairEyeColor, "Eye", "Brown")
tableSetSliceValue(HairEyeColor, "Eye", "Brown", complement=TRUE)

tableSetSliceValue(HairEyeColor, c("Eye","Hair"), c("Brown","Black"))
tableSetSliceValue(HairEyeColor, c("Eye","Hair"), c("Brown","Black"), complement=TRUE)