Ensure output of classifier predict has aligned index

[jcrist]

When returning pandas/cudf outputs, we want the index of the output to be aligned with the index of the input. This PR plumbs through index in decode_labels, adds a test for all classifiers, then applies the necessary fixes and plumbing so the output has an aligned index.

In the long run I hope to move index alignment handling to the reflect decorator so method code doesn't need to worry about this, but for now this is the most idiomatic and cleanest solution.

Split out and expanded from #8039.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 173b16b8-8849-49d1-b397-903643dc79cd

📥 Commits

Reviewing files that changed from the base of the PR and between 4c30f68 and a8f9394.

📒 Files selected for processing (10)

• python/cuml/cuml/common/classification.py
• python/cuml/cuml/ensemble/randomforestclassifier.py
• python/cuml/cuml/fil/fil.pyx
• python/cuml/cuml/linear_model/logistic_regression.py
• python/cuml/cuml/linear_model/mbsgd_classifier.py
• python/cuml/cuml/multiclass/multiclass.py
• python/cuml/cuml/neighbors/kneighbors_classifier.pyx
• python/cuml/cuml/svm/linear_svc.py
• python/cuml/cuml/svm/svc.py
• python/cuml/tests/test_base.py

---

📝 Walkthrough

Summary by CodeRabbit

• Bug Fixes
  • Classification model predictions now consistently preserve the index information when using pandas DataFrames as input across all classifiers, including RandomForest, LogisticRegression, KNeighborsClassifier, SVC, and LinearSVC.
  • Index metadata is properly maintained throughout the prediction pipeline for both single-target and multi-target classification scenarios.

Walkthrough

This PR enhances the classification prediction pipeline to preserve and propagate index metadata through prediction outputs. The decode_labels function now accepts an optional index parameter, and multiple classifiers are refactored to capture index information from intermediate results and pass it through to final outputs.

Changes

Index Propagation Through Classification Pipeline

Layer / File(s)	Summary
API Enhancement python/cuml/cuml/common/classification.py	decode_labels gains an optional index=None parameter and now wraps all output paths (multi-target, single-target numeric, and non-numeric) with CumlArray(..., index=index) or DataFrame index=index to preserve the provided index.
Core Infrastructure python/cuml/cuml/fil/fil.pyx	ForestInference.predict returns CumlArray(data=result, index=proba.index) for classifier probabilities instead of a plain array, enabling index propagation upstream.
Multiclass Base Implementation python/cuml/cuml/multiclass/multiclass.py	_BaseMulticlassClassifier.predict and decision_function now call check_inputs(return_index=True) to capture index and return CumlArray(data=out, index=index), establishing index propagation for multiclass estimators.
Classifier Predict Methods python/cuml/cuml/ensemble/randomforestclassifier.py, python/cuml/cuml/linear_model/logistic_regression.py, python/cuml/cuml/linear_model/mbsgd_classifier.py, python/cuml/cuml/svm/linear_svc.py, python/cuml/cuml/svm/svc.py, python/cuml/cuml/neighbors/kneighbors_classifier.pyx	Each classifier's predict method is refactored to defer to_output("cupy") conversion, capture index from intermediate results (decision_function, predict_proba, or FIL output), and pass the captured index to decode_labels. SVC additionally updates predict_proba to preserve index throughout the probability computation path.
Test Coverage python/cuml/tests/test_base.py	Added verification that predictions preserve the input DataFrame's index when using pandas output type, ensuring index propagation is correctly maintained end-to-end.

---

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~50 minutes

---

Possibly related PRs

• rapidsai/cuml#8023: Modifies prediction output handling and index propagation in ForestInference and RandomForestClassifier, with decode_labels enhancements to preserve and return indexes alongside predictions.

---

Suggested labels

Cython / Python, improvement, non-breaking

---

Suggested reviewers

• hcho3
• dantegd

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title 'Ensure output of classifier predict has aligned index' clearly and specifically describes the main objective of the changeset - ensuring that classifier predictions maintain aligned indices with input data.
Description check	✅ Passed	The description directly relates to the changeset by explaining the purpose of plumbing index through decode_labels, adding tests, and applying fixes for index alignment in classifier predict outputs.
Docstring Coverage	✅ Passed	Docstring coverage is 90.00% which is sufficient. The required threshold is 80.00%.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[csadorf]

I agree that we should try to lift some of the churn into the reflect decorator in the future.

LGTM!

[jcrist]

/merge