Fix uplift tree p-value NaN from division by zero (#585)

[jeongyoonlee]

Summary

• Fixes uplift p_value is NAN #585
• When a tree node has zero treatment or control observations (n_t=0 or n_c=0), the p-value variance formula p_t*(1-p_t)/n_t + p_c*(1-p_c)/n_c divides by zero, producing NaN
• Also guards against zero variance (when p=0 or p=1 for all observations in a group)
• Returns p_value=1.0 (maximally non-significant) in these degenerate cases, which is the correct statistical interpretation — no evidence of treatment effect with no observations

Test plan

• pytest tests/test_uplift_trees.py — 23 passed

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR fixes issue #585 where UpliftTreeClassifier produced NaN p-values due to division by zero in the Wald test formula when a tree node had zero treatment or control observations (n_t=0 or n_c=0), or when variance was zero (p=0 or p=1 for all observations in a group). The fix returns p_value=1.0 (maximally non-significant) in these degenerate cases.

Changes:

• Guards the p-value computation against division by zero (n_t=0 or n_c=0) and zero variance, returning p_value=1.0 as a conservative fallback
• Modifies .gitignore to narrow the .claude/ ignore pattern and add .worktrees/ entries (appears unrelated to the fix)

Reviewed changes

Copilot reviewed 1 out of 2 changed files in this pull request and generated 1 comment.

File	Description
causalml/inference/tree/uplift.pyx	Adds guards around the p-value Wald test computation to prevent NaN from division by zero or zero variance
.gitignore	Changes .claude/ ignore to .claude/.worktrees/ and adds duplicate .worktrees/ entries

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review. Take the survey.

[jeongyoonlee]

Addressed the review comment: added test_uplift_tree_pvalue_no_nan_with_sparse_groups in 144e6b6. It uses a heavily imbalanced dataset (45 control, 2 treatment1, 3 treatment2) with min_samples_treatment=0 and min_samples_leaf=1 to force zero-count nodes, then asserts no NaN in predictions.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 3 changed files in this pull request and generated no new comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review. Take the survey.