🚀 Features
- Add 3D support for
ep.pp.miss_forest_impute()(#1052) @sueoglu - Add 3D support for
ep.pp.mice_forest_impute()(#1055) @sueoglu - Add stratified Table One for baseline comparisons (#1066) @Zethson
- Add CONSORT-style branching for
CohortTracker(#1077) @Zethson - Add 3D & lazy Dask array support for
ep.pp.encode()(#1078) @Zethson - Add in-house causal inference module —
ep.tl.iptw(),ep.tl.g_computation(),ep.tl.aipw(),ep.tl.propensity_score_matching(), meta-learners (ep.tl.t_learner()/ep.tl.s_learner()/ep.tl.x_learner()), andep.tl.covariate_balance()— replacing thedowhydependency (#1076) @Zethson
🐛 Bug Fixes
🧰 Maintenance
- Adjust to
ehrdata0.3.0 release, addinganndata0.13.0 support (#1084) @eroell - Make
ehrdatacompatible withanndata>=0.12.13(#1065) @eroell - Efficient, dependency-free implementation of the MCAR test (#1056) @agerardy
- Replace
EhrapyConfigwith scverse-miscSettings(#1070) @Zethson - Address pandas
FutureWarning(#1063) @agerardy - Remove unused imports, debug prints, and stale TODOs (#1075) @Zethson
- Fix invalid rst roles in
installation.mdleiden section (#1073) @Zethson - Improve fate tutorial robustness (#1081) @eroell
- Cache CI datasets and harden against flaky downloads (#1080) @sueoglu
- Mirror scanpy's image-comparison tolerances to fix CPU plotting test failures (#1083) @sueoglu
- Unpin numpy in run_notebooks workflow (#1085) @eroell
💥 Breaking changes
Full Changelog: v0.14.0...v0.15.0