type_variable: Functions for defining the types of variables

Description Usage Arguments Value Examples

Description

All functions permit fully-customised types of variable to be defined. Functions other than type_custom already include standard default values which make the definition of standard variable types easier and quicker.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
type_custom(
  type = "custom",
  init = c(0, 10),
  randomXj = function(n, v) {     rnorm(n, v[1], v[2]) },
  x2v = function(x) {     c(mean(x), sd(x)) },
  v2x = function(v) {     v[1] },
  smooth = list(quote(smooth_lin(x, xt, 1)), quote(smooth_dec(x, xt, 0.9, 10))),
  ...
)

type_real(...)

type_positive(...)

type_negative(...)

Arguments

type

label for identifying the type of variable. The name is not used internally in any case.

init

numeric vector of starting values of parameters of the sampling distribution.

randomXj

function for randomly generating variable values according to the sampling distribution. The function should take the number of observations to be generated as a first argument, and the vector of parameters as a second argument; a vector of random values should be returned.

x2v

function for updating the parameters of the sampling distribution. No smoothing is needed. The function should take a single argument to be used for updating the parameters.

v2x

function for obtaining point values of variable from the parameters of the sampling distribution.

smooth

list of unevaluated expressions of smoothing functions for each parameter of the sampling distribution.

...

further arguments to be included into the typevar object. In case of function for predefined types, it is possible to use ellipsis for overwriting default values (see ยง Examples).

Value

An object of class type and typevar, where type is the value of the argument type passed to type_custom, or predefined lables (if not overwritten) in case of other functions.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
# Define a new type of real variable where the first parameter of the
# sampling distribution is updated through the median (instead of the
# mean):
type_real(
  type = 'real2', 
  x2v = function(x) { c(median(x), sd(x)) }
)

# Define a new type of real variable whith different smoothing
# parameters:
type_real(
  type = 'real3', 
  smooth = list(
    quote(smooth_lin(x, xt, 0.8)),
    quote(smooth_dec(x, xt, 0.99, 15))
  )
)

noisyCE2 documentation built on Nov. 9, 2020, 5:13 p.m.