Model Tuner 0.0.13a

Version 0.0.13a

• Updated bootstrapper
• evaluate_bootstrap_metrics
• Added notebooks/xgb_early_bootstrap_test.py to test it
• Updated requirements.txt file for dev testing
• Fixed sampling error on low number of samples inside bootstrapper

Model Tuner 0.0.12a

• Xgboost bug fixes
• Zenodo updates
• Pickle model fixes with np import
• ADASYN and SMOTE fix with no fit happening when calibrating

model_tuner 0.0.11a

• updated readme for PyPI
• previous version not saved on setup; re-release to 0.0.11a

model_tuner 0.0.10a

• updated readme for PyPI

model tuner 0.0.09a

• number of estimators now extracted from XGBoost model object
• early stopping fixed

model_tuner 0.0.08a

AutoKerasClassifier

• Changed 'layers' key to store count instead of list to avoid exceeding MLflow's 500-char limit.
• Simplified function by removing key filtering loop.

model tuner 0.0.07a

What's Changed

• Kfold threshold tuning fix by @elemets in #17

Full Changelog: 0.0.06a...0.0.07a

0.0.06a

What's Changed

• Updating best_params: ref before assignment bug by @elemets in #16

Full Changelog: 0.0.05a...0.0.06a

model_tuner 0.0.05a

• Bootstrapper:
  • Fixed import bugs
  • Fixed Assertion bug to do with metrics not being assigned
• Early stopping:
  • Leon: fixed bug with SelectKBest and ADASYN where the wrong code chunk was being utilized
  • Arthur: Verbosity fix

model_tuner version 0.0.02a

• temporarily commented out updated apache software license string in setup.py
• updated logo resolution