What's New

A little maintenance and a small bug fix.

🐛 Bug Fixes

• get_repo() did not return repo
• Respect "method" kwarg in combining mods

🧪 Testing

• Run dvc repro to check new lyscripts

⚙️ Miscellaneous Tasks

• Bump requirements

Change

• Slightly improve logging

What's New

Just some improvements to the cos and the switch to loguru for logging. No functionality should have been altered.

📚 Documentation

• Add __repr__ & explanation to C
• Mention private attribute _column_map
• Mention execute method of Q objects
• Fix unfinished sentence in utils

Change

• In __repr__, add parentheses around combination of AndQ and OrQ.
• Switch to loguru for logging

What's New

🚀 Features

• Add central to short name columns

🐛 Bug Fixes

• & and | with None return original Q. Previously, Q(...) | None would return a query that evaluated to True everywhere.

📚 Documentation

• List defined operators on Q (&, |, ~, ==) in the docstring of CombineQMixin.

🧪 Testing

• ensure that & and | with None return original Q.

What's New

Another bug fix: Previously, sub- and superlevel involvement was only computed for columns not already present in the table. Now, it is by default computed and correctly replaces unknown values.

🚀 Features

• (utils) Add better update func for pandas

🐛 Bug Fixes

• Order of sub-/superlevel inference
• Don't ignore present sub-/superlvl cols

What's New

This release fixes a bug where completely unobserved LNLs would be reported as healthy when using the ly.combine() method. Also, This method is now roughly 20x faster than before 🚀

🐛 Bug Fixes

• If an LNL of a patient was unobserved (i.e., all diagnoses None), then the method ly.combine() returns None for that patient's LNL. Fixes #13

🧪 Testing

• Change the doctest of ly.combine() to check whether #13 was fixed.

What's New

This is a clean-up update. Some stuff I thought might be useful turned out to be unnecessary, while other things got better names. Two small features have also made it.

🚀 Features

• Can now combine Q with None to yield Q again.
• Add contains operator to C, Q objects. This calls pandas' str.contains method.

🧪 Testing

• Fix wrong name in doctests

Change

• [breaking] Add, rename, delete several methods:
  • LyDatasetConfig is now just LyDataset
  • the path property is now path_on_disk
  • the get_url() method has been removed
  • the get_description() method has been removed
  • added get_content_file() method to fetch and store remove content
  • load() was renamed to get_dataframe()
  • the repo argument was changed to repo_name
• (utils) [breaking] Rename enhance func to infer_and_combine_levels.

Remove

• [breaking] Two unused funcs for markdown processing were removed
• (load) [breaking] Drop join_datasets, since it's not needed. All it did was run pd.concat(...).

What's New

Just a quick hotfix.

🐛 Bug Fixes

• (load) Fix a bug where datasets with multiple subsites (e.g. 2024-umcg-hypopharynx-larynx) would cause an error because of a missing maxsplit=2 argument.

What's New

Small features and refactorings.

🚀 Features

• (load) add get_repo() method that fetches remote repository information for a `LyDatasetConfig
• (load) make authentication more flexible
• (utils) put sub-/superlevel inference in its own utility function

What's New

With this release, we make the switch from rapidly evolving 0.0.X versions to something that changes a little more slowly. However, we still consider the library experimental and breaking changes may still occur frequently.

🚀 Features

• (utils) Add often needed enhance function to complete sub-/superlevel involvement and infer maximum likelihood status.

🐛 Bug Fixes

• Avoid KeyError in infer_superlevels

⚙️ Miscellaneous Tasks

• Add link to release 0.0.4

Change

• infer_su(b|per)levels skips inferring involvement of sub-/super LNLs that are already present
• (load) Rename skip_disk to use_github
• (query) Rename in_ to isin for C object

What's New

🚀 Features

• [breaking] Make several helper functions private (e.g., _max_likelihood())
• (utils) Add more shortname columns, like surgery for ("patient", "#", "neck_dissection")
• (load) Allow search for datasets at different locations on disk
• (query) Add C object for easier Q creation
• (query) Add in_ to C object
• (validate) Add transform_to_lyprox function

🐛 Bug Fixes

• (load) Resolve circular import of _repo

📚 Documentation

• Add intersphinx mapping to pandera
• Expand module docstrings
• Update README.md with library examples

🧪 Testing

• Fix failure due to changing order of items in set

Change

• (validate) Add args to renamed validation
• Import useful stuff as top-level
• Make main() funcs private

Remove

• (load) [breaking] load_dataset() not needed, one can just use next(load_datasets())