Curated tabular datasets bundled with the imputify
library for examples, tests, and missing-data benchmarks. Every file is a single parquet:
feature columns first, target column last, no missing values. Column names are
snake_case; columns that the upstream sources include for identification but not
modeling (IDs, free-text names) have been dropped.
X, y = load("iris") #… See the full description on the dataset page:
https://huggingface.co/datasets/gabfssilva/imputify-datasets.
