Exercises

Week 1:

• exercises 2.3, 2.8, 2.9, 7.4, 7.5 and 7.6
  • solutions to exercises 2.3, 2.9, 7.4, 7.5 and 7.6
  • solution to exercise 2.8
     

Week 2:

• exercises 3.6, 3.13, 3.14, 3.17 (without lasso for now), 7.7 and 7.9
  • solutions to exercises 3.6, 3.13, 3.14, and 7.7
  • solution to 3.17
  • solution to exercise 7.9
     

Week 3:

• exercise 3.16, 3.19
  • solutions to exercises 3.16 and 3.19
• starting from exercise 3.17: (i) include lasso as well; (ii) add 4 columns to Table 3.3 with the results for non-negative garrote, SCAD, adaptive lasso and elastic-net (using prostate data); (iii) repeat exercise 3.17 using logistic (penalized) regression for (weighted) LS, best subset, lasso and ridge regression
  • solution to the extension of exercise 3.17 with prostate data
  • solution to exercise 3.17 with binary response

Week 4:

• exercises 6.2, 6.3 and 6.7;
  • solutions to exercises 6.2, 6.3, and 6.7
• estimate the density of the variables tobacco and alcohol in the South African Heart Disease dataset by using the classical kernel method (using both "Gaussian" and "Epanechnikov" kernels, changing the bandwidth). Repeat the estimation using the Hjort-Glad estimator. What is the advantage of using a density with only positive support (e.g., an exponential) as the initial guess?
  • solution to the additional exercise

Week 5:

• exercises 5.1, 5.3, 5.4, 5.7 and 5.9.
  • solutions to exercises 5.1, 5.7, and 5.9 (from a previous year)
  • solution to exercise 5.3
  • solution to exercise 5.4 (by a student from a previous year)

Week 6:

• see file with exercises by Odd Kolbjørnsen
  • solutions (including that of ex. 5.9 from previous week)

Week 7:

• exercise 9.1
  • solution to exercise 9.1
• apply a single classification tree and bagging to the spam data and contrast the results of the two algorithms
  • solution to extra exercise
• exercise 1 from the STK-IN4300 second mandatory assignment of 2018
  • solution to exercise 1 from 2nd 2018 mandatory assignment 

Week 8:

• exercises 8.4, 15.1, 15.4, 15.5 and 15.6;