v0.35.0

Feature

• Basic styling (9c26121)

v0.34.0

Feature

• Even prettier formatting (1cadde3)

Fix

• Prettier output for df_dimensions_diff (4ba728a)

v0.33.0

Feature

• Minor cleanups to (198bc44)

v0.32.1

Fix

• Compute correct diff when using df diff decorator (1879b93)

v0.32.0

Feature

• Allow to drop visits after exclusion timestamp (77b7fc8)
• For filters, only print dimensions if anything changed (0e5587b)

v0.31.0

Feature

• Add self-healing (1c15d46)

Fix

• Handle uninitialised wandb (1de530c)
• Move self-healing to the error location (ca295c3)
• No NA in outcome col in test data (d95b79c)
• Synth_data_generator did not add nas (ec5e3b5)
• Incorrect import (30696c2)

v0.30.2

Fix

• Dont modify the config when flattening to dict (03f50e3)
• Correctly unpack nested pydantic basemodels (1d9437f)

v0.30.1

Fix

• Add _ as prefix to when looking for outcome columns to drop (2c0ce05)
• Make exclusion timestamp optional in EvalDataset (65e1d0f)
• min_lookahead_days no longer matching multiple values (1ed5833)

v0.30.0

Feature

• Allow manual specification of relative path for config dir (d7c8de3)

Fix

• Add Optional as possible type for exclusion_timestamp (c650905)

v0.29.0

Feature

• Wrap training loop in try/except (ec8630a)
• Wandb alert on exception to main script (f53a2f9)

Fix

• Handle floats when binning continuous data (23fc939)
• Use mean imputer if imputation is required and none specified (f427a39)
• Hydra main should return metric to optimise (26600f6)
• Incorrect specificatin of logistic-regression (3137e7a)
• Drop edges if duplicate (aa32bfe)
• Create model eval dir if does not already exist (574b778)
• Guard against no wandb when getting eval dir (028feb3)
• Relative import of custom_plot_artifacts (e966005)
• New trainer path (61c17e9)
• Specify_config_dir relatively (15af2bd)
• Init wandb before adding wandb decorator (3e07440)