The benchmark's logistic regression does not converge on German Credit Lending and Healthcare Readmission
Maintainers usually reply within 1 day
Nobody has claimed this yet.
Assessment
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Newbie friendliness
- 57/100
- Issue type
- Bug
- Clarity
- Clearly specified
- Activity status
- Active
- Tech stack
- python, scikit-learn
- Domain
- machine-learning, testing-qa
Research direction
Start in faircode/models.py at MODEL_FAMILIES and inspect how build_model constructs logistic regression. Run the reproduction command in the issue, then check the German Credit Lending and Healthcare Readmission audit results under results/ and use diff_results.py to record changed rows. Done when benchmark fits emit no convergence warnings and the pull request lists the changed rows.
Written by the indexing model from the issue text.
Description
Where: faircode/models.py MODEL_FAMILIES, results/.
Current gap: Fitting the benchmark's logistic regression on the S2 features of these two audits raises a scikit-learn ConvergenceWarning: the solver stops at max_iter on unscaled features, so the coefficients, and every number built on them, depend on where it stopped.
Repro:
PYTHONPATH=. python3 -c "import warnings; from faircode.benchmark import load_audit; from faircode.models import build_model; d=load_audit('German Credit Lending'); warnings.simplefilter('error'); build_model('logistic_regression', d.random_state).fit(d.X.iloc[d.idx_train], d.y.iloc[d.idx_train])"
Suggested approach: Scale numeric features inside the estimator (a pipeline) or raise max_iter, rerun the two audits, and record the movement with diff_results.py.
Done when: no benchmark fit warns, and the changed rows are listed in the pull request.
- Dominant language
- HTML
- Stars
- 47
- Forks
- 45
- Avg merge
- 3h 8m
- Merged PRs (30d)
- 96
Getting set up
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from yakew7/Fair-Code
-
ci enhancement testing
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
Maintainers usually reply within 1 day
-
enhancement reproducibility tooling
Difficulty 2/5 1-3 hours Newbie friendliness 76/100
Maintainers usually reply within 1 day
-
enhancement reproducibility
Difficulty 2/5 1-3 hours Newbie friendliness 84/100
Maintainers usually reply within 1 day
-
community enhancement tooling
Difficulty 2/5 1-3 hours Newbie friendliness 84/100
Maintainers usually reply within 1 day
-
The benchmark OG card and the home card hard-code counts and domains that are also kept elsewhereOpenenhancement tooling
Difficulty 2/5 1-3 hours Newbie friendliness 65/100
Maintainers usually reply within 1 day
All issues in yakew7/Fair-Code
Similar issues
-
Bug in Web UI: Unknown model arch 'sd15' when selecting SD 1.5Possibly taken @ngo-binh0108 claimed this today. Open
Difficulty 1/5 Under an hour Newbie friendliness 86/100
ostris/ai-toolkit#1092 ·
Maintainers usually reply within 1 day
-
regression skipped skipped_bmg
Difficulty 2/5 1-3 hours Newbie friendliness 76/100
intel/torch-xpu-ops#5892 ·
Maintainers usually reply within 2 days
-
Difficulty 2/5 1-3 hours Newbie friendliness 64/100
containers/podman-desktop-extension-ai-lab#4853 ·
Maintainers usually reply within 1 day
-
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
Maintainers usually reply within 1 day
-
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
google-research/perch-hoplite#191 · 1 comment ·
Maintainers usually reply within 16 days