Reject complex calibrated association features

[FlorianPfaff]

Bug

CalibratedPairwiseAssociationModel validates named component planes as real numeric values, but its direct-feature predict_proba adapter used asarray(..., dtype=float64) without first checking the input dtype. Backend complex arrays can therefore lose their imaginary component during conversion and reach the classifier as different real-valued features.

For example, a feature row containing 1 + 2j can be converted to 1.0 rather than being rejected. This makes invalid association features produce plausible probabilities and costs instead of a deterministic validation error.

Fix

• route direct predict_proba feature tensors through the existing real-numeric backend validator before flattening
• preserve the existing float64 conversion for valid real-valued features
• reject invalid complex features before invoking the wrapped classifier

Regression coverage

The new test supplies a backend-native complex feature tensor, verifies that ValueError is raised, and confirms that the wrapped model's predict_proba method is never called.

Validation

• the source and regression test compile successfully in the patch workflow
• the final branch is based on current main, zero commits behind, and changes only the association-feature implementation and one focused test
• repository CI should provide the complete NumPy, JAX, PyTorch, lint, packaging, and documentation matrix

[github-actions]

✅MegaLinter analysis: Success

Descriptor	Linter	Files	Fixed	Errors	Warnings	Elapsed time
✅ COPYPASTE	jscpd	yes		no	no	29.5s
✅ JSON	prettier	7	0	0	0	1.31s
✅ JSON	v8r	7		0	0	4.22s
✅ MARKDOWN	markdownlint	68	0	0	0	2.0s
✅ MARKDOWN	markdown-table-formatter	68	0	0	0	0.86s
✅ PYTHON	black	1828	64	0	0	100.68s
✅ PYTHON	isort	1828	99	0	0	2.9s
✅ REPOSITORY	betterleaks	yes		no	no	2.41s
✅ REPOSITORY	checkov	yes		no	no	58.59s
✅ REPOSITORY	gitleaks	yes		no	no	17.97s
✅ REPOSITORY	git_diff	yes		no	no	0.37s
✅ REPOSITORY	secretlint	yes		no	no	92.62s
✅ REPOSITORY	syft	yes		no	no	6.51s
✅ REPOSITORY	trivy-sbom	yes		no	no	7.23s
✅ REPOSITORY	trufflehog	yes		no	no	33.3s
✅ YAML	prettier	11	0	0	0	0.83s
✅ YAML	v8r	11		0	0	13.48s
✅ YAML	yamllint	11		0	0	0.69s

Notices

📣 MegaLinter 9.5.0 is out! Discover the new features and security recommendations in the release announcement. (Skip this info by defining SECURITY_SUGGESTIONS: false)

See detailed reports in MegaLinter artifacts

Your project could benefit from a custom flavor, which would allow you to run only the linters you need, and thus improve runtime performances. (Skip this info by defining FLAVOR_SUGGESTIONS: false)

• Documentation: Custom Flavors
• Command: npx mega-linter-runner@9.6.0 --custom-flavor-setup --custom-flavor-linters PYTHON_BLACK,PYTHON_ISORT,COPYPASTE_JSCPD,JSON_V8R,JSON_PRETTIER,MARKDOWN_MARKDOWNLINT,MARKDOWN_MARKDOWN_TABLE_FORMATTER,REPOSITORY_CHECKOV,REPOSITORY_GIT_DIFF,REPOSITORY_GITLEAKS,REPOSITORY_BETTERLEAKS,REPOSITORY_SECRETLINT,REPOSITORY_SYFT,REPOSITORY_TRIVY_SBOM,REPOSITORY_TRUFFLEHOG,YAML_PRETTIER,YAML_YAMLLINT,YAML_V8R

Show us your support by starring ⭐ the repository