v0.28.0

Feature

• Add support for custom columns in create eval dataset (294eeac)
• Check for custom col names when loading (c031850)
• Add custom columns to configs (14dc05e)
• Migrate n_hba1c to custom_artifact (51b2095)

Fix

• Convert bins to primitive before adding to plot (13748d6)
• Add guards for unspecified custom_columns (333dc5a)

v0.27.2

Fix

• Avoid overflow when drawing heatmap (0b6876b)
• Disable wandb upload in tests (4dfba79)
• Update configs to remove unused keys (d6a1950)
• Missing imports (e3928eb)
• Only one return statements (194da42)
• Clean up imports (abc7170)

v0.27.1

Fix

• No reaosn to call main twice (93e961c)

Documentation

• Specify trainerspec output type (9f6f29f)

v0.27.0

Feature

• Add suggested cols if missing using Levenshtein edit distance (ae6a6c2)
• Unify wandb group naming (7ae1bb0)
• Check that specified col names exist in dataset (7208e13)
• Add wandb alert on exception (ef3df84)
• Update project name (aaaf52f)
• Add model-training suffix to trainer (f2ac4dc)

Fix

• Imputation method is not required (f47d8b1)
• Only check col names in col schema (ee61991)
• Sleep between workers to avoid segfaults (fcd7cd2)
• Rename preprocessing keys (eee7768)
• Incorrect paths (ef4e925)
• Broken imports (cb361e5)
• Post_split config addressing (f76dd5a)
• Broken imports (9a1531c)
• Broken imports in train_model (7852599)

Documentation

• Improve docs (ad0b842)

v0.26.1

Fix

• Missing imports (6ac4032)

v0.26.0

Feature

• Type hints (544569c)
• Set min for y-axis on performance_by_whatev (144542b)

Documentation

• Adding optional (c94a1f5)
• Adding optional (b07ff52)

v0.25.0

Feature

• Set min,max for y_axis on performance plots (6381ca7)
• Log which features remain after feature selection (51c4746)

Fix

• Removed redundant attribute checks (3427844)

Documentation

• Updated name in docstring (2b75ab1)
• Output type hint and docstring edit (4d9d505)

v0.24.0

Feature

• Expand diff decorator to rows (daf0c45)

v0.23.0

Feature

• Print diff for each processing step (83c4689)

Fix

• Allow none as imputation method (af0127c)

v0.22.0

Feature

• Cache dataset processing (d2eac6e)
• Use mutual_info_classif (5c650d3)

Fix

• Convert to tuple before hashing (9bd67c0)