sebp/scikit-survival

Implements specialized models that account for censored data—where event outcomes are partially unknown—enabling time-to-event predictions in clinical and reliability domains. Integrates seamlessly with scikit-learn's preprocessing, cross-validation, and pipeline infrastructure while supporting both uncensored and right-censored observations. Provides multiple survival model variants optimized through convex solvers (ECOS, OSQP) with dependencies on NumPy, SciPy, and pandas for numerical computation.

1,282 stars and 277,459 monthly downloads. Used by 5 other packages. Actively maintained with 6 commits in the last 30 days. Available on PyPI.