Skip to contents

Benchmark Numeric Imputation Methods Using Relative Squared Error (RSE)

Source: R/imputation_diagnostics.R
tune_impute_numeric.Rd

Evaluates multiple numeric imputation methods (mean, median, and KNN) by artificially masking observed values and computing Relative Squared Error (RSE) and MAE.

Usage

tune_impute_numeric(
  data,
  target_cols,
  methods = c("mean", "median", "knn"),
  prop_mask = 0.15,
  seed = 123,
  checkpoint_dir = NULL,
  shard = NULL,
  checkpoint_mismatch = c("recompute", "overwrite", "error")
)

Arguments

data

A data frame with numeric columns.

target_cols

Character vector of numeric columns to evaluate.

methods

Imputation methods to benchmark: `"mean"`, `"median"`, and/or `"knn"`.

prop_mask

Proportion of observed values to artificially mask (default 0.15).

seed

Random seed for reproducible masking.

checkpoint_dir

Optional directory path where per-task checkpoints are stored.

shard

Optional shard specification (e.g. `c(1, 2)`) for distributed computation.

checkpoint_mismatch

Action on mismatch: `"recompute"`, `"overwrite"`, or `"error"`.

Value

A tibble ranking the imputation methods by RSE and MAE across variables.