The final assignment for this class is a multi-week project. The project is self-driven but the expectation is that you will work in groups to demonstrate your ability to do something original with your newfound pythonic abilities.

It’s up to you, but some suggestions include:

• Testing ideas towards a Capstone project
• Replicating and extending analysis done in a published paper
• Working with an existing codebase/model to apply an interesting ML method
• Performing a novel analysis on a dataset

Expectations

• This should involve original work from your team (size of group: anywhere between 1-38)
• Level of effort should be ~2-3 weeks worth of work
• Submission will be a repository including:
  • Code: your own and perhaps from an existing project
  • Documentation:
    • Overview of the problem
    • Description of the dataset you used (input features, outcome, dimensions, etc)
    • How to run the code (dependencies, etc.)
    • Decisions made along the way, including trade-offs (e.g., cut X for time so our solution may lack Y)
    • Example output (what does it do?)
    • Citations (data, code, papers)
  • Short-form presentation (slides and notes or video, not during lecture):
    • 10 minutes with fellow classmates as the target audience
    • Problem statement
    • Existing work you pulled from
    • Your contribution
    • Tools/methods used
    • Issues overcome along the way
    • PPT or document style OK

Inspiration

Data & code

• The Incredible PyTorch

  List of papers, code, examples using PyTorch

• Kaggle

  ML competition/collaboration site

• Keras code examples

  Official examples of implementation using Google’s TensorFlow Keras

• PLOS papers with available data

  Searching PLOS for keywords likely to have available data, refine further to get topics interesting to you

• PLOS recommended repositories (data, code, and sometimes both)

  Lots here, mostly data repositories

• The Pudding

  Visual essays with data