Midae: R6 class for Midae imputer

In agnesdeng/misle: Multiple imputation through statistical learning

R6 class for Midae imputer

Description

Set up a Midae imputer object with specified hyperparameters and then obtain multiple imputed datasets

Methods

Public methods

• Midae$new()

• Midae$train()

• Midae$impute()

• Midae$impute.new()

• Midae$clone()

---

Method new()

Usage

Midae$new(
  data,
  n_h = 2L,
  encoder_structure = c(128L, 64L, 32L),
  decoder_structure = "reversed",
  learn_rate = 1e-04,
  iteration = 400,
  batch_size = 500,
  input_drop = 0.2,
  hidden_drop = 0,
  act = tf$nn$elu,
  weight_decay = "default",
  epsilon = 1e-07,
  rho = 0.95,
  decay = 0.9,
  momentum = 0.9,
  optimizer = "Adam"
)

---

Method train()

Usage

Midae$train(verbose = TRUE, print_freq = 1)

---

Method impute()

Usage

Midae$impute(
  m = 5,
  onehot = FALSE,
  all.numeric = FALSE,
  add.noise = FALSE,
  SD = 1,
  pmm = FALSE
)

---

Method impute.new()

Usage

Midae$impute.new(
  newdata = data,
  m = 5,
  onehot = FALSE,
  all.numeric = FALSE,
  pmm = FALSE
)

---

Method clone()

The objects of this class are cloneable with this method.

Usage

Midae$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

agnesdeng/misle documentation built on Sept. 22, 2023, 8:48 p.m.