obs_gen: Observation generation

In koohyun-kwon/HTEBandSim: What the Package Does (One Line, Title Case)

Description

Observation generation

Usage

obs_gen(
  n,
  n.sim,
  reg.name = c("AK", "AK.bin"),
  reg.spec,
  M,
  cvar.spec,
  dist,
  scale,
  x.spec = c("unif", "beta"),
  x.l = -1,
  x.u = 1
)

Arguments

n	number of observations
n.sim	number of simulation to use.
reg.name	class of regression functions to be used; "AK" denotes regression functions in Section 5 of Armstrong and Kolesár (2020); "AK.bin" denotes denotes the case where y = 1(f(x) + u > 0), using regression functions in Section 5 of Armstrong and Kolesár (2020), where u is a normal random variable.
reg.spec	a number denoting the type of regression function among the class of functions chosen in reg.name.
M	bound on the second derivative
cvar.spec	specification for conditional variance function, one of c("homo", "het.AK").
dist	distribution for noise function; choose among normal ("norm"), log-normal distribution ("lnorm") or logistic ("logis").
scale	scale parameter, a positive constant multiplied to the noise.
x.spec	distribution of the regressor, either uniform distribution ("unif") or normalized beta(2, 2) distribution ("beta"), both with the support of [-1, 1].
x.l	lower end of the support of the regressor.
x.u	upper end of the support of the regressor.

Value

list of following components

x

vector of regressors

y

matrix of outcome variables with nrow = n and ncol = n.sim

Examples

obs_gen(100, 100, "AK.bin", 2, 1, "het.AK", "norm", 1/2, "beta")

koohyun-kwon/HTEBandSim documentation built on Dec. 21, 2021, 7:43 a.m.