Impute missing values via random forest (missForest)

Wraps missForest to impute numeric columns using non-parametric random forests. Requires the missForest package to be installed (Suggests). Non-numeric columns are untouched. If no numeric columns contain NAs, data is returned unchanged.

Usage

impute_missforest(data, ntree = 100, cols = NULL, verbose = FALSE, ...)

Arguments

data

A data.frame or tibble containing missing values.

ntree

Integer; number of trees to grow in each forest (passed to missForest). Default 100.

cols

Optional character vector of numeric columns to impute. Defaults to all numeric columns with at least one NA.

verbose

Logical; if TRUE, prints progress and a completion summary. Default FALSE.

...

Additional arguments passed to missForest::missForest().

Value

A data.frame/tibble with numeric columns imputed by missForest (or mean fallback).

Details

Notes:

• missForest uses iterative RF training; it can be slow on wide/high-NA data.

• Errors (e.g., insufficient unique values) are caught and a deterministic mean imputation fallback is applied to the selected numeric columns.

References

Stekhoven DJ, Buhlmann P (2012). “MissForest—non-parametric missing value imputation for mixed-type data.” Bioinformatics, 28(1), 112–118. doi:10.1093/bioinformatics/btr597 .

Examples

# \donttest{
if (requireNamespace("missForest", quietly = TRUE)) {
  df <- tibble::tibble(a = c(1, NA, 3), b = c(2, 4, NA), c = 5)
  impute_missforest(df, ntree = 50, verbose = TRUE)
}
#> impute_missforest(): preparing inputs (3 rows, 2 columns)
#> Warning: impute_missforest(): missForest failed; falling back to mean imputation.
#> Warning: impute_missing(): column 'a' has high missingness (33.3%).
#> Warning: impute_missing(): column 'b' has high missingness (33.3%).
#> impute_missforest(): results: fallback mean imputation; imputed 2 values across 2 columns.
#> # A tibble: 3 × 3
#>       a     b     c
#>   <dbl> <dbl> <dbl>
#> 1     1     2     5
#> 2     2     4     5
#> 3     3     3     5
# }