make_synthetic_feature_table

Note

This page is a reference documentation. It only explains the function signature, and not how to use it. Please refer to the Habitat Guide and Python API guide (v2.0) for usage.

make_synthetic_feature_table(n_rows: int = 60, n_features: int = 12, task: Literal['binary', 'survival'] = 'binary', rng: int | Generator | SeedSequence = 0) → FeatureTable[source]

Build a deterministic tabular dataset for fast ML golden tests.

A single signal feature separates the endpoint classes or correlates with survival; the remaining columns are pure noise so selectors and classifiers have a stable correct answer to recover.

Parameters:

  • n_rows – Number of subjects / rows.

  • n_features – Total number of model-input columns including signal.

  • 1. (n_features must be at least)

  • task – "binary" for a single label column or "survival" for follow-up time plus an event indicator.

  • rng – Seed controlling feature noise and endpoint draws.

Returns:

A FeatureTable with explicit outcome semantics.

Examples

>>> from habit.datasets import make_synthetic_feature_table
>>> table = make_synthetic_feature_table(n_rows=10, n_features=4, rng=42)
>>> table.frame.shape
(10, 6)
>>> list(table.feature_columns)
['signal', 'noise0', 'noise1', 'noise2']
>>> table.outcome.task
'binary'