0.3.0

Deepchecks Version 0.3.0

Supported Environments

• Support running from non-jupyter environments - Suite outputs can be exported to html, and Check or Suite outputs can be exported to JSON for serialization. HTML export: #638, JSON export: #665
• Added support also for non-sklearn models. Supporting any model object that implements the predict method (and also predict_proba for classification), see supported models #594
• Fixed dependency warning when installing on google colab #577
• Fixed links from inside suite and check output to open in a new tab

Display and Output Improvements

• Output can be exported to html #638
• Suite output is now in a widget, enabling to browse quickly between checks, with and without conditions or display #610
• Distribution graphs smoothened, graphs looks better also when feature has outlier values #664
• Improved error handling. Checks that didn’t run due to irrelevant input (e.g. no index column) are no longer displayed as errors. #640

Docs

Improved documentation for many check examples:

• Trust Score
• Train Test Feature Drift
• Whole Dataset Drift
• Simple Model Comparison
• Single Feature Contribution
• Data Duplicates

Added documentation for export types:

• To HTML and To JSON

New Contributors

• @idow09 made their first contribution 🥳
• @BioGeek made their first contribution 🥳
• @shaypal5 made their first contribution 🥳

Maintainers

@matanper , @ItayGabbay , @yromanyshyn , @JKL98ISR , @benisraeldan , @deepchecks-bot , @shir22 , @nirhutnik , @noamzbr