Hello there!
Thank you for the checking out neuropredict, and for your interest in contributing.
This is a relatively new project (less than an year old) 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 Tests
- write tests for command line usage
- write tests for input format support
- write tests for integration among different modules
- write unit tests or property tests to assert its advertised functionality
- testing delivery/installation system to ensure it works as advertised on several platforms
- More documentation
- write examples for each type of usage
- write more docs on results produced
- improve docs on how to use the results
- write docs on how not to use the results
- Portability
- Implement support for more input formats (besides CSVs, pyradigm, arff)
- 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)
- 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
- building an API-like interface for programmatic access for advanced users
- improving code base in terms of modularity
- improving the executing speed by optimizing the code
- improving the speed by offering GPU support
- ideas for GUI development
- ideas for integration with other use cases
- Any other features or idea that you think are interesting and helpful o you or your friends.
Thanks for your time and help.
Under construction
To be improved and completed.