Skip to content

Conversation

@hxrshxz
Copy link

@hxrshxz hxrshxz commented Nov 26, 2025

Metadata

Item Value
Reference Issue NA
New Tests Added Yes
Documentation Updated No
Change Log Entry Add integration test for regression task split integrity (Task 2280)

Description

What does this PR implement?

  • Adds a new integration test tests/test_tasks/test_regression_task_splits.py.
  • The test validates the split logic of a real regression task (OpenML task 2280):
    1. Confirms the task type is SUPERVISED_REGRESSION.
    2. Checks that the task uses 10‑fold cross‑validation with a single repeat.
    3. Verifies that train and test indices are NumPy arrays, non‑overlapping, and within dataset bounds.
    4. Ensures the union of all test indices covers the entire dataset.

Why is this change necessary?

  • Existing tests cover classification and clustering tasks but lack regression split verification.
  • Correct split handling is essential for reproducible experiments and for downstream UI components that consume OpenML data.

How to run the test

# From the repository root (virtual environment activated)
pytest tests/test_tasks/test_regression_task_splits.py

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant