Pitch a Project!

Project Pitches

Please pitch a project idea in the #project-pitches channel on Campuswire. Your project idea will not necessarily be the final project you work on — indeed, many of you will work on someone else’s idea. That’s ok! I am still expecting everyone to make a pitch.

To Pitch a Project:

First, read the course description of the project. Then, think about what you’d like to do!

Second, write your pitch! To write your pitch:

  1. Write one or two sentences about the big picture: what problem you’d like to address and how machine learning fits in.
  2. If your project needs data (almost all will), state your data plan. There are three ways:
    1. Describe a data set to which you currently have access.
    2. Link to an online data set that is suitable for addressing your problem (please note: your link must be neither from Kaggle nor the UCI Machine Learning Repository).
    3. Describe a specific approach by which you will collect your own data.
  3. State what kind of problem your pitch involves. Is it a classification problem, a regression problem (predicting a quantitative outcome rather than a qualitative label), or an unsupervised problem like clustering?
  4. Describe how you’ll judge whether your project is successful. What are you looking to have produced/achieved by the end of the semester?
  5. Close your pitch by letting us know: why are you excited about this topic?

Where Can I Look For Data?

There’s no general guidance here, but there are a few approaches:

  1. Many scholarly papers release data sets as companions to their papers. If you have a topic in mind, you can search for papers on that topic and see if they have released data.
  2. It is ok to browse Kaggle for interesting data sets, but you must use a data set which has an original source outside of Kaggle, and must use that original version.
  3. Some sites run by governments, nonprofits, and NGOs host data sets. For example, data.gov offers a hub for many data sets. The EPA also has a data portal which includes APIs.

How To Submit Your Pitch

Beyond posting the pitch to Campuswire, there’s nothing else that you need to do. We’ll check your pitch off when we see it on Campuswire and log credit for you in Gradescope.