Simulate data from array.

Simulate data (slice of) an array: Simulate n observations from the array x conditional on the variables in margin (a vector of indices) takes values given by margin.value

simulateArray(x, nsim = 1, margin, value.margin, seed = NULL)

# S3 method for class 'table'
simulate(object, nsim = 1, seed = NULL, margin, value.margin, ...)

# S3 method for class 'xtabs'
simulate(object, nsim = 1, seed = NULL, margin, value.margin, ...)

# S3 method for class 'array'
simulate(object, nsim = 1, seed = NULL, margin, value.margin, ...)

Arguments

x, object

An array.

nsim

Number of cases to simulate.

margin, value.margin

Specification of slice of array to simulate from.

seed

Seed to be used for random number generation.

...

Additional arguments, currently not used.

Value

A matrix.

Note

The current implementation is fragile in the sense that it is not checked that the input argument x is an array.

Author

Søren Højsgaard, sorenh@math.aau.dk

Examples

## 2x2 array
x <- parray(c("a", "b"), levels=c(2, 2), values=1:4)

x <- tabNew(c("a", "b"), levels=c(2, 2), values=1:4)

## Simulate from entire array
s <- simulateArray(x, 1000)
xtabs(~., as.data.frame(s))
#>    b
#> a     1   2
#>   1  95 297
#>   2 198 410

## Simulate from slice defined by that dimension 1 is fixed at level 2
s <-simulateArray(x, 1000, margin=1, value.margin=2)
xtabs(~., as.data.frame(s))
#> b
#>   1   2 
#> 340 660 

## 2 x 2 x 2 array
x <- parray(c("a", "b", "c"), levels=c(2, 2, 2), values=1:8)
## Simulate from entire array
s <-simulateArray(x, 36000)
xtabs(~., as.data.frame(s))
#> , , c = 1
#> 
#>    b
#> a      1    2
#>   1  998 3008
#>   2 1940 4065
#> 
#> , , c = 2
#> 
#>    b
#> a      1    2
#>   1 5044 7015
#>   2 5919 8011
#> 

## Simulate from slice defined by that dimension 3 is fixed at level 1
s <-simulateArray(x, 10000, 3, 1)
xtabs(~., as.data.frame(s))
#>    b
#> a      1    2
#>   1  949 2994
#>   2 1956 4101