Changelog

v0.1.1 (2026/03/12)

Refactored

• Split large modules (config.py, pipeline.py, evaluation/data_quality.py) into organized subpackages.

• Reorganized visualization into domain-specific visualizers with a unified Visualizer interface.

• Improved structure and readability of multistate analysis, stability analysis, grid search, and CLI validation code.

• Removed decorative separator comments from source and test files.

Tests

• Reorganized tests to mirror the project structure and added shared fixtures.

• Expanded test coverage.

Fixed

• Minor fixes in encoder, preprocessing, external validation, and feature selection discovered during testing.

v0.1.0 (2026/03/04)

Initial release.

• LCA/LPA via StepMix with automatic model selection and configurable cluster-size constraints

• Train/test split before preprocessing; full-cohort refit after selection

• Outcome analysis (logistic regression, chi-square/Fisher), survival analysis (Cox PH, log-rank), and multistate modelling

• Consensus clustering stability and internal validation

• Feature characterisation with effect sizes (Cohen’s d, Cramer’s V)

• Interactive HTML report, forest plots, KM/NA curves, state occupation plots