Transformation tools

Tools for transforming between N(0,1) variables and other distributions in predictor expressions

Usage

bru_forward_transformation(qfun, x, ..., tail.split. = 0)

bru_inverse_transformation(pfun, x, ..., tail.split. = NULL)

Arguments

qfun

A quantile function object, such as qexp

x

Values to be transformed

...

Distribution parameters passed on to the qfun and pfun functions

tail.split.

For x-values larger than tail.split., upper quantile calculations are used internally, and for smaller values lower quantile calculations are used. This can avoid lack of accuracy in the distribution tails. If NULL, forward calculations split at 0, and inverse calculations use lower tails only, potentially losing accuracy in the upper tails.

pfun

A CDF function object, such as pexp

Value

• For bru_forward_transformation, a numeric vector

• For bru_inverse_transformation, a numeric vector

Examples

u <- rnorm(5, 0, 1)
y <- bru_forward_transformation(qexp, u, rate = 2)
v <- bru_inverse_transformation(pexp, y, rate = 2)
rbind(u, y, v)
#>          [,1]        [,2]        [,3]      [,4]       [,5]
#> u -1.65033134 -0.97938791 -1.10782537 0.6109105 -0.4277990
#> y  0.02535073  0.08938046  0.07191707 0.6535024  0.2035322
#> v -1.65033134 -0.97938791 -1.10782537 0.6109105 -0.4277990