Exercises
Practise the following exercises to get a firm grasp on the concepts learned so far:
- Did you notice that I put CV in italics when I said that using
k=27seems like a safe bet as measured by the minimization of the CV error? Did you wonder why? I (quite deliberately) made a gaffe in choosing the k in the k-NN from Figure 10.4. My choice wasn't wrong, per se, but my choice of k may have been informed by data that should have been unavailable to me. How might have I committed a common but serious error in hyper-parameter tuning? How might I have done things differently?
- Remember that we spent a long time talking about the assumptions of linear regression? In contrast, we spent virtually no time discussing the assumptions of logistic regression. Although logistic regression has less stringent assumptions than its cousin, it is not assumption-free. Think about what some assumptions of logistic regression might be. Confirm your suspicions by doing research on the web. My omission of the...
