Exploratory Project

Neural networks are one of the most broadly applicable machine learning tools we’ve seen in the course, and can be used in a wide variety of different settings. Instead of completing Traffic, you have the option for this unit to complete an exploratory project, where you build a neural network for a classification task of your choosing.

To complete the exploratory project, you should:

  • Pick a topic area of interest to you: options include art, biology, business, energy, engineering, epidemiology, finance, government, language, law, medicine, music, sports, and more.
  • Find a data set about the topic.
  • Send a brief proposal email to Brian, Doug, and your assigned teaching fellow describing what topic you’d like to explore and what data set you plan to use.
  • Once your project is approved (unapproved projects may not be submitted!), build a neural network using TensorFlow to perform a classification task based on your data set.
    • The expectation is that the bulk of your time on this project should be spent on implementing this neural network, not manipulating data. Projects that seem to not satisfy this expectation will have a maximum possible Design score of 3 out of 5.
  • Submit the code for your network along with a comprehensive README.md file that describes, in one or more paragraphs, how to run your program. Your teaching fellow should, ideally, be able to do everything they would need to do with the project by reading the README alone.
  • In addition, in your README, document (in at least a paragraph or two) your experimentation process. What did you try? What worked well? What didn’t work well? What did you notice?

Data sets of interest might include

How to Submit

  1. Download a ZIP file containing your implementation of this project.
  2. Go to CSCI E-80’s Gradescope page.
  3. Click Project 5: Exploratory.
  4. Drag and drop your ZIP file to the area that says “Drag & Drop”.
  5. Click Upload.

You should see a message that says “Project 5: Exploratory submitted successfully!” Contact your teaching fellow if not!