Benchmarks

MultivariateImputer is benchmarked on five datasets and two synthetic missingness patterns. The charts summarize the headline scores; the table below them carries every metric of every run.

Results

Methodology

Each run takes a dataset, drops rows with existing missing values (to preserve ground truth), injects synthetic missingness, imputes, and scores only the masked entries.

Missingness patterns

• MAR 10% — 10% of cells removed at random across the whole table.

• Blocks 20% × 30% — contiguous blocks covering 20% of the rows in 30% of the columns.

Datasets

Three numeric-only scikit-learn tabular datasets (Diabetes, Wine, Breast Cancer), plus mixed-type Titanic and a synthetic mixed dataset combining numeric and categorical columns.

Metrics

Metrics are split by data type: regression metrics (RMSE, MAE, R², MAPE, SMAPE, median AE, bias, normalized RMSE) for numeric columns, and classification metrics (accuracy, balanced accuracy, macro precision/recall/F1, MCC, Cohen’s κ) for categorical columns. Coverage reports the fraction of masked values that received finite predictions.