Day 5

Today

• • Class discussion of warmup project

• Review of questions on Chapters 10 and 11

  • Finish "machine learning from the bottom up"

• Launch of the "Choose Your Own Adventure Project"

For Next Time

• "Finish" machine learning from the bottom up.

• Writeup your Project Proposal and Learning Goals for the CYOA project.

Class Discussion of Warmup Project

As a class we'll be discussing the Warmup Project.  Here is the relevant text from the project writeup.

You should come to class ready to engage in a class-wide discussion.  Prompts for discussion include:

• • 1. The various models you tried

    2. The various resources you found that helped you along the way

    3. Tools that seem to be especially useful

    4. Processes that you tried that seemed to work well for writing good data science code.  Good code in this context could mean code that is concise, readable, maintainable, extensible, etc.  You are not expected to have all the answers!  Even if you tried something that seems suboptimal, put it out there, and see if someone has suggestions for improvement.  Topics could include version control strategies, workflows, code refactoring, object-oriented design, etc.

    5. Reflections on strategies for learning that you tried that didn't pan out

    6. Any mysterious results that you can't yet explain.

    7. Your burning questions (have at least two open-ended discussion questions you can use to get the others in the group talking).

We will spend about twenty-five minutes discussing this as a class.

Review of Chapter 10 and 11 and Discussion of Course Feedback

I'll be going over the most common confusions on the exercises for chapters 10 and 11.

Finish "machine learning from the bottom up"

We'll be making the jump from single variable gradient descent to multivariable gradient descent.  There is a notebook in the repository under day05 that will serve as a template.  Contrasts from last time:

1. self-paced

2. emphasis on thinking before writing code (write equations on the board, write pseudo-code, etc.)

3. some additional extensions / improvements to gradient descent are available for those that want to explore some algorithms in more depth

You may not get through the whole activity during class, and given the fairly light homework load for next class, I encourage you to finish it up on your own time. You will learn a lot!

If you are a fan of interactive debuggers, you should consider using the ipdb module as a way to debug your code.