em_mixedgc: EM algorithm to fit Gaussian copula

In udellgroup/mixedgcImp: Missing value imputation for mixed data through Gaussian copula

EM algorithm to fit Gaussian copula

Description

fit the Gaussian copula model from incomplete mixed data

Usage

em_mixedgc(
  Z,
  Lower,
  Upper,
  d_index,
  dcat_index = NULL,
  cat_input = NULL,
  start = NULL,
  trunc_method = "Iterative",
  n_sample = 5000,
  n_update = 1,
  maxit = 50,
  eps = 0.01,
  verbose = FALSE,
  runiter = 0,
  corr_min_eigen = 0.01,
  scale_to_corr = TRUE
)

Arguments

Z	Transformed latent matrix
Lower	Lower boundary of truncated intervals
Upper	Upper boundary of truncated intervals
d_index	Boolean vector with TRUE at ordinal dimensions
dcat_index	Boolean vector with TRUE at categorical dimensions
cat_input	Input for categorical dimensions
start	Initial value of copula correlation
trunc_method	Method for evaluating truncated normal moments: 'Iterative' or 'Sampling'.
n_sample	Number of MC samples, only used when trunc_method is 'Sampling'
n_update	The number of updates, only used when trunc_method is 'Iterative'
maxit	Maximum number of iterations
eps	Convergence threshold
verbose	Whether to print progress information
runiter	When set as a positive integer, the algorithm will run the specified number of iterations exactly.
corr_min_eigen	If the minimal eigenvalue of a correlation estimate is below corr_min_eigen, it will be regularized to have minimal eigenvalue equal to corr_min_eigen
scale_to_corr	Whether to scale a covariance into a correlation matrix in each EM iteration. For development purpose. Use with caution.

Value

A list containing fitted copula correlation matrix, the likelihood(objective function), Z matrix with updated ordinal entries and a complete imputed Z matrix.

corr

Fitted copula correlation matrix

loglik

The log-likelihood achieved during iteration.

Z

Incomplete Z with approximated observed ordinal mean

Zimp

Complete Z with observed entries the same as Zobs and missing entries imputed

udellgroup/mixedgcImp documentation built on Jan. 25, 2023, 7:55 p.m.