Multi Profile Viewer

Hi everyone! We’ve now released whylogs v0.5.2 🚀.
Whylogs is an open standard for data and ML logging created by WhyLabs. 👩🏽‍🔬

This version includes:

✨ Features

• Select Multiple Profile in viewer #272 [@valer-whylabs]

🐛 Bug Fixes

• Remove default metadata writer #269 [@lalmei]

v0.5.1 Unicode release

Hi everyone! We’ve now released whylogs 0.5.1 🚀.
whylogs is an open standard for data and ML logging created by WhyLabs. 👩🏽‍🔬

This version includes:

• 😊 Unicode support for stringtracker,
• 🐛 Various bug fixes.
• 📚 Documentation updates

📚 Full documentation can be found at https://docs.whylabs.ai or simply head to our Github https://github.com/whylabs/whylogs to learn more., We are always open to feedback at our repo or slack http://join.slack.whylabs.ai/.

✨ Features

• Unicode support for string tracker #266 [@lalmei]

🐛 Bug Fixes

• Check additional attribute in Logger #260 [@lalmei]
• Datasetprofiles reads and writes with smart_open #262 [@lalmei]
• Use default config only if config is None #267 [@lalmei]
• Handle non-string labels when serializing confusion matrix #265 [@andyndang]
• String tracker update with custom character list #263 [@lalmei]

📚 Documentation

• Update Coverage link #261 [@lalmei]

v0.5.0 NLP metrics and Auto Segmentation Release

whylogs release v0.5.0

Hi everyone! We’ve now released whylogs v0.5.0 🚀. whylogs is an open standard for data and ML logging created by WhyLabs. 👩🏽‍🔬

This release includes:
📝 Support for tracking NLP metrics such as Word Error Rate (WER), Match Error Rate (MER), Word Information Lost (WIL)
𝌋 Auto segmentation to help you discover meaningful segments in your dataset. Save segments for future use so that you can create separate profiles for each segment as data is streamed during deployment or training. Compare segments to help identify bias, and so much more!
🐛 Various Bug fixes.
📚 Documentation updates: Full documentation can be found at https://docs.whylabs.ai or Github https://github.com/whylabs/whylogs
📈 You can push your whylogs profiles directly to WhyLabs Platform for monitoring if you have an API key. Need an API key? Sign up for the free account here: https://bit.ly/whylabs-free-sign-up
❤️ Give us feedback or chat with our team on Slack: http://join.slack.whylabs.ai/

Below is the detailed change log:

✨ Features

• API and method to auto segment features in a model. #255 [@bernease, @lalmei]
• NLP metrics such as Word Error Rate #248 [@lalmei]

🐛 Bug Fixes

• Fix non-tracked character positions in String Tracker merge#257 [@lalmei]
• Fix number of tokens in StringTracker merge #256 [@lalmei]
• Add example showing custom tokenizer for string tracking #253 [@lalmei]

📚 Documentation

• Add String Tracker Notebook example link in README #252 [@lalmei]
• Add example showing custom tokenizer for string tracking #253 [@lalmei]
• update release draft to use correct version number #249 [@lalmei]

🔨 Maintenance

• Refactor datasetprofile's flatten profile #250 [@lalmei]
• Fix make bump-x scripts #258 [@lalmei ]

whylogs: Release v0.4.9

Hi everyone! We’ve now released whylogs v0.4.9 🚀. This release includes support for tracking and visualizing string features, along with some bug fixes and documentation updates. Whylogs is an open standard for data and ML logging created by WhyLabs. 👩🏽‍🔬

This version includes:

✨ Features

• String features such as length, word count, and character position tracking #236 [@lalmei]
• String related visualizations for character position tracking in visualization library #236 [@lalmei]
• Protobuf are written in serialized delimited format by default #243 [@cswarth]
  Note: To read previously default written files use datasetProfile.read_protobuf(protobuf_file_path, delimited_file=False)

🐛 Bug Fixes

• Store null_count in schema tracker only #234 [@andyndang]
• Feature list in the static HTML profile viewer dynamically updates based on checked inferred types #230 [@valer-whylabs]

📚 Documentation

• Added Content profile viewer #239 [@valer-whylabs]
• Notebook example for string tracking #236 [@lalmei]

v0.4.8

v0.4.8 Profile Viewer Release

✨ Features

• Static HTML Profile Viewer #227 [@valer-whylabs]
• Add profile viewer method #229 [@lalmei]
• Enable publishing to WhyLabs #222 [@andyndang]

📚 Documentation

• update README with visualization methods #229 [@lalmei]
• ROV Logging Example #221, #224 [@lalmei]
• Add PR template and Contribution guide #220 , #223 [@lalmei]
• Fix README in pypi #219 [@andyndang]

v0.4.5 Regression

✨ Features

• Move profile IO into background threads #204 [@andyndang]
• accept magnitude and units in with_rotation_time param #180 [@cswarth]
• Regression metrics #171 [@lalmei]
• python 3.9 support #170 [@lalmei]

🐛 Bug Fixes

• Add none case classification metrics merge #189 [@lalmei]
• Create a new logger if the existing one is closed #209 [@andyndang]
• Enable custom number of histograms #165 [@andyndang]
• make sure merger doesn't affect model type #191 [@lalmei]
• Handle None case in Confusion matrix merging #163 [@andyndang]
• Return 0 for a single value tracker #162 [@andyndang]
• Fall back to number unique summary or string unique summary #164 [@andyndang]

📚 Documentation

• Update Spark example #218 [@andyndang]
• add example section #217 [@lalmei]
• Update README.md #215 [@lalmei]
• Configuration and docs updates #203 [@naddeoa]
• fix typos in logger doc strings #178 [@cswarth]
• Fixed sphinx documentation #166 [@andyndang]

🎨 Improve structure and format of code

• move example notebooks from example repo #147 [@lalmei]
• Add poetry to manage dependencies #200 [@naddeoa]
• Reformat the codebase with black, update format targets #201 [@naddeoa]

🚨 Testing

• test jupyter notebooks in repo #147 [@lalmei]

v0.3.2 🤹‍♀️ Fixes and 🤡 Patches

🐛 Bug Fixes

• Fix merge logic for various corner cases #159 [@andyndang]
• Default ModelProfile to None #159 [@andyndang]
• Fixed merging two ModelProfile after deserializing from protobuf bytes #159 [@andyndang]
• Fix constraints bug AttributeError: 'ValueConstraint' object has no attribute 'update' #158 [@cswarth ]

Metrics and Constraints Release

✨ Features

• sketch model metrics for classification #153 [@lalmei]
• create. and apply constraints to datasets #152 [@cswarth]
• bounding boxes data tracking #134 [@lalmei]
• track full local datasets #134 [@lalmei]

🐛 Bug Fixes

• refactor track() -> update() #158 [@cswarth]
• add python version reqs #136 [@lalmei]

📚 Documentation

• 📚 added code of conduct #142 [@lalmei]

⬆️ Dependencies

• Migrate to datasketches 2.2.0b1 #151 [@andyndang]
• Bump cryptography from 3.2.1 to 3.3.2 #150 [@dependabot]
• Update outdated requirements #141 [@lalmei]

⬇️ Dependencies

• remove a dependency on bleach #144 [@lalmei]
• Remove s3fs refs and use smart open instead #140 [@lalmei]

🚨 Testing

• 🚀 added release drafter #143 [@lalmei]

✨ Images and Mlflow release

whylogs v0.1.13

✨ Features

• Image data support, create sketches of:

  • Images metadata
  • Image data features
  • Custom image feature transformations
  • see image example

• Segment support

  • Create segmented profiles from your structured data

• Add MLFLOW support

  • Start creating data profile during your mlflow steps.
  • see upcoming python examples

• Set log rotation with time.

  • Automatically create new profiles every second, minute, hour, or day at any set interval

🚨 Add more coverage tests

• Coverage s3 with mock s3
• Log rotation with freeze gun to mock time traveling.
• Added coverage tracking
• Added maintainability tracking with code climate.

🐛 Numerous Bug Fixes:

• Bug with tracking None datatypes
• Bug with s3 writers duplicating folder structure locally
• Pass a custom config path
• Various fixes

➕ Dependencies

• upgrading various decencies
• adding optional image loading dependencies

Release v0.1.2

• MLFlow integration