Validation against R

Future validation should compare QAPLogit against R sna::netlogit.

The current repository includes a skipped placeholder test under:

tests/validation_against_r/test_qap_against_sna_netlogit.py

Required alignment

Validation should explicitly align:

• directed versus undirected dyad handling;
• diagonal inclusion or exclusion;
• predictor matrix ordering;
• permutation count;
• random seed or exported permutation indices;
• coefficient naming;
• empirical p-value convention;
• tolerance for floating-point comparisons.

Backend note

Use the statsmodels backend for R validation.

The sklearn backend is useful for faster regularized permutation workflows, but it does not expose standard errors, z-values, or model-based p-values.

Current status

R validation fixtures are pending.

Until fixture export is added, the Python implementation is covered by:

pytest tests/unit/test_qap.py tests/unit/test_qap_permutation.py -q