Homework 4

Contents

Homework 4#

Deadline

Due Sunday, March 2nd, 2025 at midnight on Crowdmark.

Instructions#

This homework covers four classes. Problems listed below are from the textbook.

  • Fri 2/14, we covered 5.1.1 - 5.1.2

  • Mon 2/17, we covered 5.1.3

  • Wed 2/19, we covered 5.1.4

  • Fri 2/21, we covered 5.1.5

These problems don’t quite line up to individual classes, so I will just list them out below.

  • 5.4.3

    • Note: “Explain” means no code is necessary. You can use words, pictures, and/or pseudocode to show me that you understand how this procedure works.

  • 5.4.5: The Default data set is on the DataSets page

    • Hint: For part (c), you do not need to regenerate the data. You have the same data set, but you are generating a new split of the data. This could be as easy as setting a new seed.

  • 5.4.8 (a-e)

    • Added part f: Repeat part (c) using \(k\)-fold CV for \(k=5,10,15,20\). Plot your results for error vs. degree for all these plus the LOOCV version. What do you notice?

    • You don’t need to plot this for the training error, it’s annoyingly difficult to get that out of the easy-mode version of \(k\)-fold CV. Just do it for test error. If you really want to try to get the training error plotted, too, take a look here.

Important

Standard instructions for submissions and deadlines can be found on the Homework Info Page.

Contents