simulate_mixture_cube: Simulate a mixture cube to test 'CVtreeMLE' against simulated...

In blind-contours/CVtreeMLE: Cross Validated Decision Trees with Targeted Maximum Likelihood Estimation

Simulate a mixture cube to test CVtreeMLE against simulated ground-truth.

Description

Simulate a mixture cube. This creates three correlated mixture variables that are associated with two confounders W1 and W2. First mixtures are generated from a multivariate normal. A multinomial outcome is generated based on betas input for W1 and W2 - associating each W with a part of the mixture cube. In each part of the mixture cube, transform the multivariate normal mixture to a uniform distribution, respecting the bounds for parts of the cube. A three variable cube with one threshold per variable has 8 subspaces. An outcome is then generated as a linear combination of different subspaces.

Usage

simulate_mixture_cube(
  n_obs = 500,
  splits = c(0.99, 2, 2.5),
  mins = c(0, 0, 0),
  maxs = c(3, 4, 5),
  mu = c(0, 0, 0),
  sigma = matrix(c(1, 0.5, 0.8, 0.5, 1, 0.7, 0.8, 0.7, 1), nrow = 3, ncol = 3),
  w1_betas = c(0, 0.01, 0.03, 0.06, 0.1, 0.05, 0.2, 0.04),
  w2_betas = c(0, 0.04, 0.01, 0.07, 0.15, 0.1, 0.1, 0.04),
  mix_subspace_betas = c(0, 0.08, 0.05, 0.01, 0.05, 0.033, 0.07, 0.09),
  subspace_assoc_strength_betas = c(1, 1, 1, 1, 1, 1, 1, 7),
  marginal_impact_betas = c(0, 0, 0),
  eps_sd = 0.01,
  binary = FALSE
)

Arguments

n_obs	Number of observations for which to generate data
splits	Vector indicating where thresholds should be placed for each mixture variable
mins	Vector indicating the minimum values for each mixture variable
maxs	Vector indicating the maximum values for each mixture variable
mu	Vector indicating the mean values for each mixture variable
sigma	Matrix of the variance-covariance structure used to generate the mixture variables
w1_betas	Vector of betas that define the subspace probability relationship with covariate W1
w2_betas	Vector of betas that define the subspace probability relationship with covariate W2
mix_subspace_betas	Vector of betas that define the subspace probabilities
subspace_assoc_strength_betas	The outcome Y generated by each partition of the mixture cube
marginal_impact_betas	Vector of betas that define the marginal impact each mixture variable has
eps_sd	Random error included in the generation of Y
binary	TRUE/FALSE depending on if the outcome should be binary

Value

obs: A data frame of the simulated data for the mixture cube.

blind-contours/CVtreeMLE documentation built on June 22, 2024, 8:53 p.m.