Assemble Imputed Columns From Their Best-`mtry` Runs

Given a sweep produced by [missforest_sweep_mtry()] and a table naming the winning `mtry` for each column, pulls each column out of the run that won it. Different columns may come from different runs.

Usage

missforest_impute_by_mtry(sweep, best)

Arguments

sweep

A named list of [missforest_oob_by_mtry()] results, named by `mtry` as a character string (as [missforest_sweep_mtry()] builds it).

best

A data frame with one row per column, containing at least the columns `column` and `mtry` – typically the `best` element returned by [missforest_sweep_mtry()].

Value

A tibble of imputed columns, one per row of `best`.

Details

Runs are looked up by **name** (`sweep[[as.character(mtry)]]`), not by position. Positional lookup happens to work only when the swept `mtry` values are exactly `1:n`; it silently returns the wrong run for any other sweep, such as one starting above 1 or skipping values.

See also

[missforest_sweep_mtry()]