Hello there!
Thank you for the checking out neuropredict, and for your interest in contributing.
This is an interdisciplinary project with an exciting roadmap, both in scope and in long term goals. The core functions of neuropredict as a tool (learning, cross-validation and visualization) have been implemented already in their basic form. However, these could be improved significantly in various ways, which include:
- More documentation
- write more examples for each type of usage
- write more docs on results produced
- improve docs on how to use the results and API
- write docs on how not to use the results
- Portability
- Propose and implement web-friendly output formats for results e.g.
- json output compared to pickle files for numerical outputs
- java script based visualizations instead of PDFs
- interface to run neuropredict over web or on a browser (like a service)
- Implement support for more input formats (besides CSVs, pyradigm, arff)
- Propose and implement web-friendly output formats for results e.g.
- New features
- propose interesting and new use cases that neuropredict could support to further multiply its value
- integrate Keras to offer support for CNNs and other variants
- Also, take a look at open issues
- Better software engineering
- improve the API for programmatic access for advanced users
- improving the speed by offering GPU support, and/or more internal optimization
- ideas for GUI development
- ideas for integration with other use cases
- More Tests
- write unit tests or property tests to assert its advertised functionality
- testing delivery/installation systems to ensure it works as advertised on Windows and other platforms
- Any other features or idea that you think are interesting and helpful o you or your friends.
Thanks for your time and help, and being a champion of open source!