Skip to contents

MISL: Multiple Imputation by Super Learning (v2.0)

Source: R/misl.R
misl.Rd

Imputes missing values using multiple imputation by super learning.

Usage

misl(
  dataset,
  m = 5,
  maxit = 5,
  seed = NA,
  con_method = c("glm", "rand_forest", "boost_tree"),
  bin_method = c("glm", "rand_forest", "boost_tree"),
  cat_method = c("rand_forest", "boost_tree"),
  ord_method = c("polr", "rand_forest", "boost_tree"),
  cv_folds = 5,
  ignore_predictors = NA,
  quiet = TRUE
)

Arguments

dataset

A dataframe or matrix containing the incomplete data. Missing values are represented with NA.

m

The number of multiply imputed datasets to create. Default 5.

maxit

The number of iterations per imputed dataset. Default 5.

seed

Integer seed for reproducibility, or NA to skip. Default NA.

con_method

Character vector of learner IDs, a list of parsnip model specs, or a mixed list of both, for continuous columns. Default c("glm", "rand_forest", "boost_tree").

bin_method

Character vector of learner IDs, a list of parsnip model specs, or a mixed list of both, for binary columns (values must be 0/1/NA or a two-level factor). Default c("glm", "rand_forest", "boost_tree").

cat_method

Character vector of learner IDs, a list of parsnip model specs, or a mixed list of both, for unordered categorical columns. Default c("rand_forest", "boost_tree").

ord_method

Character vector of learner IDs, a list of parsnip model specs, or a mixed list of both, for ordered categorical columns. Default c("polr", "rand_forest", "boost_tree").

cv_folds

Integer number of cross-validation folds used when stacking multiple learners. Reducing this (e.g. to 3) speeds up computation at a small cost to ensemble accuracy. Default 5. Ignored when only a single learner is supplied.

ignore_predictors

Character vector of column names to exclude as predictors. Default NA.

quiet

Suppress console progress messages. Default TRUE.

Value

A list of m named lists, each with:

datasets

A fully imputed tibble.

trace

A long-format tibble of mean/sd trace statistics per iteration, for convergence inspection.

Details

Built-in named learners (see list_learners()):

  • "glm" - base R (logistic for binary, linear for continuous)

  • "rand_forest" - ranger

  • "boost_tree" - xgboost

  • "mars" - earth

  • "multinom_reg" - nnet (unordered categorical only)

  • "polr" - MASS (ordered categorical only)

Any parsnip-compatible model spec can also be passed directly via the *_method arguments. Named strings and parsnip specs can be mixed in the same list:


library(parsnip)
misl(data,
  con_method = list(
    "glm",
    rand_forest(trees = 500) |> set_engine("ranger")
  )
)

The mode (regression vs classification) is always enforced by misl regardless of what is set on the spec.

Parallelism

Imputation across the m datasets is parallelised via future.apply. To enable parallel execution, set a future plan before calling misl():


library(future)
plan(multisession, workers = 4)
result <- misl(data, m = 5)
plan(sequential)

Examples

# Using named learners (same as v1.0)
set.seed(1)
n <- 100
demo_data <- data.frame(x1 = rnorm(n), x2 = rnorm(n), y = rnorm(n))
demo_data[sample(n, 10), "y"] <- NA
misl_imp <- misl(demo_data, m = 2, maxit = 2, con_method = "glm")

# Using a custom parsnip spec
if (FALSE) { # \dontrun{
library(parsnip)
misl_imp <- misl(
  demo_data, m = 2, maxit = 2,
  con_method = list(
    "glm",
    rand_forest(trees = 500) |> set_engine("ranger")
  )
)
} # }