kGOFTest: kGOFTest

In vcoscrato/ktest: Estimated kernel densities comparisons

Description

Performs a hypothesis test for goodness-of-fit based on the estimated kernel densities.

Usage

kGOFTest(data, rfunc, dfunc, perm = TRUE, B = 5000,
  bw = bw.nrd0(data), npoints = 512, threads = detectCores() - 1,
  param_names = NULL)

Arguments

data	A numeric vector of data.
rfunc	A function to generate data from the real density (see examples).
dfunc	A function to evaluate real density values (see examples).
perm	Boolean indicating weather to obtain the p-value trough the permutation test or just return return the common area between densities.
B	Number of permutations.
bw	The bandwidth used to estimate the kernel densities.
npoints	The number of points used to estimate the kernel densities.
threads	Number of cores to be used for parallel computing.
param_names	A vector of variable names (as character). This parameter can be ignored when threads = 1. When using more then 1 threads, it is needed to export the global parameters name on the rfunc and dfunc functions (see examples).

Value

A list containing:

- density: The estimated density for the data.

- dfunc: The theoric density function.

- ca: Common area between the kernel and the theoric density.

- pvalue: The p-value generated by the permutation test (if perm = TRUE).

Examples

#When using no extra parameters on rfunc and dfunc:

data = rnorm(100)

rfunc = function(n) {
  return(rnorm(n, 0, 1))
}

dfunc = function(x) {
  return(dnorm(x, 0, 1))
}

kGOFTest(data, rfunc, dfunc)

#When using parameters on rfunc and dfunc:

data = rnorm(100)

param1 = 0

param2 = 1

var_names = c(param1, param2)

rfunc = function(n) {
  return(rnorm(n, param1, param2))
}

dfunc = function(x) {
  return(dnorm(x, param1, param2))
}

kGOFTest(data, rfunc, dfunc, param_names = c('param1', 'param2'))

vcoscrato/ktest documentation built on May 9, 2019, 10:01 p.m.