Read chapter 4 in ISLR through the end of section 4.3. Read from the middle of p. 147 through to the end of section 4.4.3. Read section 4.6.2, the lab section for logistic regression. Please read section 4.3.4 especially carefully (it's only a couple of pages

1. The value of the logistic function for input 0 is:

a) 0 b) 0.5 c) 1

2. In Table 4.2, the result of logistic regression seems to show that if a person is a student, the probability of default on a loan increases (the coefficient for student[Yes] is positive). However, in Table 4.3, the result of logistic regression with additional predictors seems to show that if a person is a student, the probability of default on a load decreases. What's the correct explanation?

a) in both cases, whether or not a person is a student is not statistically associated with default on a loan, so the coefficient values are meaningless b) small coefficient values like 0.4049 and -0.6468 are both small, so the difference isn't important c) overall, students are slightly more likely to default than non-students, but when compared to other people with the same credit card balance and income, students are less likely to default

3. Assume you have about 10K training examples. Which would you expect to take longer to train: a) a kNN classifier, or b) a logistic regression classifier?

4. (Yes/No) In fitting a logistic regression model, do we seek to minimize mean squared error, as we do with a linear regression model?

5. What would be a reason for using logistic regression rather than kNN for classification? (select all that apply)

a) concern over available storage b) concern over the time needed to train c) concern over the time needed to make predictions

6. With logistic regression, is it possible that removing predictors could lead to a reduced test error rate? If you think the answer is yes, respond with a page number in the text that illustrates this. If you think the answer is no, respond with number 0.

7. (True/False) When performing logistic regression with R, you should convert predictors of type factor to numeric predictors before calling the glm() function.

8. Which is usually higher: the test error rate or the training error rate? a) test b) training

9. The output of a logistic regression model on a test example is:

a) a value representing the probability that the example belongs to class 0 b) a value representing the probability that the example belongs to class 1 c) a value that should not be interpreted as a probability

10. In some classification problems (especially medical ones) we refer to the "positive" and "negative" classes. For example: "you tested positive for heart disease". In such contexts, what is a "false positive"?

a) an incorrect prediction of the positive class

b) an incorrect prediction of the negative class

c) a correct prediction of the negative class?