sclr_ideal_data: Generate ideal data for the scaled logit model

In khvorov45/sclr: Scaled Logistic Regression

Description

Allows variation of all parameters and the creation of an arbitrary number of covariates.

Usage

sclr_ideal_data(
  n = 1000,
  theta = 0,
  beta_0 = -5,
  covariate_list = list(logHI = list(gen_fun = function(n) rnorm(n, 2, 2), true_par =
    2)),
  outcome_name = "status",
  seed = NULL,
  attach_true_vals = FALSE,
  attach_seed = FALSE
)

Arguments

n	Number of observations.
theta	Baseline risk parameter on the logit scale.
beta_0	Intercept of the linear part.
covariate_list	A list in the form of name = list(gen_fun, true_par) where gen_fun is a function that takes n as an argument and returns a vector of observations, true_par is the true parameter value of that covariate. See examples.
outcome_name	Name to give to the outcome
seed	Seed to set. If NULL, no seed will be set.
attach_true_vals, attach_seed	Whether to attach additional attributes.

Value

A tibble.

Examples

# One titre
one_titre <- sclr_ideal_data(
  covariate_list = list(
    logHI = list(gen_fun = function(n) rnorm(n, 2, 2), true_par = 2)
  )
)
sclr(status ~ logHI, one_titre) # Verify

# Two titres
two_titre <- sclr_ideal_data(
  covariate_list = list(
    logHI = list(gen_fun = function(n) rnorm(n, 2, 2), true_par = 2),
    logNI = list(gen_fun = function(n) rnorm(n, 2, 2), true_par = 1)
  )
)
sclr(status ~ logHI + logNI, two_titre) # Verify

khvorov45/sclr documentation built on March 4, 2020, 7:17 p.m.