Given a matrix of n x d-dimensional random vectors, possibly containing missing elements, estimates the mean and covariance of the best fitting multivariate normal distribution.
fit.mvn(
data,
init_mean = NULL,
fix_mean = FALSE,
init_cov = NULL,
maxit = 100,
eps = 1e-06,
report = TRUE
)
Numeric data matrix.
Optional initial mean vector.
Fix the mean to its starting value? Must initialize.
Optional initial covariance matrix.
Maximum number of EM iterations.
Minimum acceptable increment in the EM objective.
Report fitting progress?
List containing the final `Mean`, `Covariance`, and EM `Objective`. If missing data are present, a `Completed` data matrix is also returned.