Status

This project is new and experimental. It works great for local development, but I wouldn’t recommend using it for production just yet. You can follow the progress on Twitter @transitive_bs

TODO

• MVP
  • e2e working example for MinMaxScaler
  • e2e working example for TSNE
  • e2e working example for Kmeans
  • add support for class attributes
  • generate all sklearn classes
  • refactor generated code into namespaces / folders
  • validate python scikit-learn version
  • generate readme files for each directory
  • port changes to python-bridge…
  • validate generated code via tscG
  • refactor into packages
  • test build via CI
  • basic readme w/ usage and examples
  • publish sklearn package to NPM
  • contact scikit-learn for feedback
• current
  • docs
    • improve python => TS markdown and type edge cases
    • fix relative markdown links
• post-MVP
  • add support for sklearn functions (in addition to classes)
  • add support for accessing the built-in datasets
  • add support for better python exception / error handling
  • add support for positional arguments (in addition to the default keyword-based arguments)
  • add basic benchmarks
  • add python <> TS unit tests
  • add support for concurrent python requests and multiple python bridges
    • this may work currently, but it’s untested and is likely to have race conditions
  • consider changing identifier names from python snake_case to JS camelCase
  • update fork of python-bridge to be more modern
    • update deps
    • replace bluebird with native promises
    • fix python warnings
    • etc
• ideas
  • add support for polars and/or danfo.js dataframe formatting
  • add support for pandas
  • explore more efficient serialization formats for IPC
    • explore memory mapping arrays between node.js and python for efficiency
    • would need the two serialized array formats to be byte-equivalent between python and node.js which seems difficult…
  • explore pyodide as a possible alternative to the Python bridge