Due: Tuesday Feb 21, 2023 (by 2:00pm PT)
Instructions:
• The homework has a total of 40 points. The TAs will pick one problem to grade and this
problem is worth 30 points (you will get 30 points if your answers are correct or almost
correct). The remaining 10 points will be graded on completion of this assignment.
• There will be two separate submissions: one for your R code and one for your writeup.
Please submit both on Gradescope (more details for the submission of the R part are given
in “Applied questions”).
• Please follow the policy stated in the syllabus about academic integrity.
• You must read, understand, agree and sign the integrity pledge
(https://academicintegrity.ucsd.edu/take-action/promote-integrity/faculty/excel-with-integrity-
pledge.pdf) before completing any assignment for ECON178. Please include your signed
pledge in the submission of your assignment on Gradescope.
Conceptual questions
• In the textbook: 5.4.3b(ii) (p.198); 6.8.4 (p.260) parts a,b,c,d only (Recall that RSS is the
“Sum of Squared Residuals”, training RSS is the RSS for the training set and test RSS is the
RSS for the test set). Please find these questions in “textbook.pdf” under “Modules ->
Assignments”.
• Explain why predictions based on the leave-one-out CV (LOOCV) tend to be less biased but
more variable than predictions based on the K−fold CV.
Applied questions (with the use of R)
• In the textbook: 3.7.14 (p.125)
• Please find “HW_2_AppliedQuestion_2.pdf” under Modules -> Assignments, and complete
the exercises in “HW_2_AppliedQuestion_2.pdf”.