impute_em: impute_em
In joon3216/funpark: Functions written by Junkyu Park

Description Usage Arguments Value Author(s) See Also Examples

Impute missing components in X using the EM algorithm, assuming each row is a sample of multivariate normal distribution.

1	impute_em(X, max_iter = 3000, eps = 1e-08)

`X`	A data or a matrix with or without `NA`'s.
`max_iter`	A natural number; the number of iterations. `3000` by default.
`eps`	A small positive number; 1e-08 by default.

A list:

mu: The resulting mean vector estimate. If X (input) has no missing components, then this is the MLE.
Sigma: The resulting variance matrix estimate. If X (input) has no missing components, then this is the MLE.
X_imputed: An imputed data.
C: !is.na(X), where X is the input.
iter: The number of iterations.

Junkyu Park

simulate_na, Imputing missing data using EM algorithm

set.seed(1024)
mu <- c(1, 2, 6)
Sigma <- matrix(c(118, 62, 44, 62, 49, 17, 44, 17, 21), nrow = 3)
n <- 400
X_truth <- MASS::mvrnorm(n = n, mu = mu, Sigma = Sigma)
result <- simulate_na(X_truth, na_rate = .4)
result_imputed <- impute_em(result$X)