builtin/

Classic datasets, shipped with the library.

Real data available offline and by name, so an example, a benchmark or an agent's first request needs no download and no file path. This is what makes {"source": "iris"} work anywhere TuiML takes a data spec.

Datasets

Classification: iris, breast_cancer, glass, diabetes

and others.

Regression: cpu, airline.
Other: supermarket (association rule mining), reuters (text).

Layout

The ARFF files ship under builtin/data/; catalog holds the metadata table used to look a dataset up by name or by task.

Notes

Loaders return a Dataset, which also unpacks as X, y — so both styles below are the same call.

python

>>> from tuiml.datasets import load_iris, list_datasets
>>> data = load_iris()
>>> data.X.shape
(150, 4)
>>> X, y = load_iris()          # unpacks directly
>>> len(set(y.tolist()))
3
>>> "iris" in list_datasets("classification")
True

Modules

catalog

Catalog of the datasets bundled with TuiML....

0 classes · 4 functions

classification

Convenience loaders for the bundled classification datasets....

0 classes · 11 functions

regression

Convenience loaders for the bundled regression datasets....

0 classes · 1 function