Note: This assignment will look very different than our other assignments. We recommend that you start this assignment early to make sure you get all the basic setup out of the way first. Using a new course tool (Kaggle in this case) can sometimes be tricky and prone to technical difficulties on your end.

Submission

THA 5 - ML Practice on Kaggle

Initial Submission by Tuesday 05/16 at 11:59 pm.

Kaggle StarterCode Programming Concept

For each homework assignment, there will usually be two things to submit:

  • A Conceptual portion that asks you to solve conceptual questions about that week’s materials. This part counts towards the Concept Portion of your assignment grade. You will turn in Concept portions on Gradescope.
  • A Programming portion that asks you to answer questions or do an analysis involving programming. This counts towards your Programming portion of your final grade. New for this assignment: You will turn in Programming portions on Gradescope. To submit the notebook on Gradescope, you should download the .ipynb notebook from Colab and upload to the upload assignment.

Conceptual

Submit

Submit the conceptual questions directly on Gradescope. You can submit your answers as many times as you want before the late cutoff (submitting after the due date will cost late days). Conceptual questions are graded manually and feedback will be posted after a period for the course staff to grade.

Kaggle/Programming

Submit code Kaggle Submission Starter Code

Note: On Kaggle it lists the close date as two days after the due date listed here. This is intended since Kaggle does not support late submissions. The homework and your submission on Kaggle is due by the due date listed here, but you may use late days and turn it in late.

This assignment is different in that we will also be using Kaggle to evaluate the models you train. You can find the full instructions for assignment on Kaggle (link above)

As a brief summary of what you need to submit:

  • Your notebook for training the edX student model should be submitted on Gradescope. This should include your report answers as text/markdown cells in the notebook explaining the relevant portions of the code or answering any questions posed.
  • The model predictions on the test data should be submitted on Kaggle.
  • We strongly recommend using the Google Colaboratory starter code linked here.