ISYE 6414

Spring 2022

PRACTICE FINAL EXAM PART 1 SOLUTIONS

PART 1: TRUE/FALSE
1. False [3.11. Assumptions and Diagnostics] If constant variance or normality assumptions do
not hold, we apply a Box-Cox transformation to the response variable.
2. False [3.13. Model Evaluation and Multicollinearity] Multicollinearity means there is a
dependency between predicting variables which would equate to the columns in the design
matrix.
3. False [4.5. Statistical Inference]
• When R-squared is used as explained variability: The denominator of the ratio can be
thought of as the total variability in the dependent variable, or how much y varies from its
mean. The numerator of the ratio can be thought of as the variability in the dependent
variable that is predicted by the model. Thus, this ratio is the proportion of the total
variability explained by the model.
• In logistic regression, the response variable is binary. So, the classic R-squared does
notmake sense in measuring explained variation.
4. False [Topic 4.4 Lesson 15: Poisson Regression: Model Description & Estimation] + [Topic
4.2 Lesson 5: Logistic Regression: Statistical Inference]
• Interpretation of the regression coefficients of Poisson regression is in terms of log ratio of
the rate.
• Interpretation of the regression coefficients of Logistic regression is in terms of log odds.
5. True [Topic 4.5 Lesson 17: Poisson Regression: Statistical Inference]
In Poisson regression, if the sample size is small, the statistical inference is not reliable.
Thus,the hypothesis testing procedure will have a probability of type I error larger than
the significance level.
6. False [5.7. Regularized Regression: Approaches] Because LASSO can eliminate predicting
variables using the penalty while Ridge and Elastic Net retain coefficients, LASSO will have
the same number or LESS predicting variables.
7. False [5.12. Variable Selection] R-squared is not compared during stepwise variable
selection. Variables are selected if they reduce the AIC or BIC of a model.
8. True [5.3. Prediction Risk Estimate] Adding more variables will increase the variability and
possibly induce multicollinearity. Adding more variables also reduces the bias in the model
 since it has an additional predictor to conform to which keeps the model from favoring one of
the original predictors.
9. False [5.4. Model Search] We desire the model that has the smallest AIC or BIC. In addition,
in forward regression, we add (accept) variables; in backwards regression, we remove
variables from the full model. We never add or remove variables when AIC/BIC is greater
than the current model, whether we are adding or removing variables. We always want to
have a lower AIC/BIC to make a step, or we should neither remove nor add variables to the
current model.
10. True [5.3. Prediction Risk Estimate] BIC penalizes complexity more than other approaches.
11. False [5.3. Prediction Risk Estimate] For linear regression under normality, the variance
used in the Mallow's Cp penalty is the estimated variance from the full model.
12. True [5.4. Model Search]” Stepwise regression is a greedy search algorithm. It does not
guarantee to find the model with the best score.”
13. True [5.7. Regularized Regression: Approaches] Ridge regression has been developed to
correct for the impact of multicollinearity. If there is multicollinearity in the model, all
predicting variables are considered to be included in the model, but ridge regression will
allow for re-weighting the regression coefficients in a way that those corresponding to
correlated predictor variables share their explanatory power and thus minimizing the impact
of multicollinearity on the estimation and statistical inference of the regression coefficients.
14. False [5.7. Regularized Regression: Approaches] Elastic net often outperforms the lasso in
terms of prediction accuracy. The difference between the lasso and the elastic net is the
addition of a penalty just like the one used in ridge regression. By considering both
penalties, L1 and L2 together, we have the advantages of both lasso and ridge regression.
15. False [4.9. Classification] Random sampling is computationally more expensive than the K-
fold cross-validation, with no clear advantage in terms of the accuracy of the estimation
classification error rate. K fold cross-validation is preferred at least from a computation
standpoint.

PART 2: MULTIPLE CHOICE

16. B: -0.948; t-stat = (Coefficient-comparison value)/SE = (41.33160-50)/9.13907 = -0.948
17. B: One; [3.3.11: Assumptions and Diagnostics] There are 27 observations hence the rule of
thumb is 4/n = 4/27 = 0.15. From the graph, there is only one observation with a Cook's
distance measurement >0.15.
18. C: Increase of 20.284; [Topic 3.2 Lesson 6: Inference for Regression Parameters] The
estimated coefficient for doserate is 20.284.
19. C: 3, 23;
• F(p, n-p-1) = F(3, 23)
• n = # of observations=27
• p = # of predicting variables=3
• p=3
• n-p-1 = 27-3-1=23
20. C: 2.252; [Topic 4.4 Lesson 15: Poisson Regression: Model Description & Estimation]

•
21. D: Decrease by 1.21%; [Topic 4.4 Lesson 15: Poisson Regression: Model Description &
Estimation] The estimated coefficient for langarts is -0.012152. A one unit increase in
langarts gives us exp(−0.012152) = 0.9879215. It is interpreted as the expected number of
days missed decreasing by 1.21% (1-0.9879). Hence, given that the other predictors in the
model are held fixed, one unit increase in langarts results in the expected number of days
missed decreasing by 1.21%.
22. A: 6.8773; [Topic 4.4 Lesson 15: Poisson Regression: Model Description & Estimation]

23. D: 3; [Topic 4.4 Lesson 15: Poisson Regression: Model Description & Estimation] The
coefficients of intercept, langarts, and male are statistically significant at 𝛼=0.05.

24. B: Chi-Squared, 312; [Topic 4.5 Lesson 20: Poisson Regression: Goodness of Fit
Assessment Data Example] Hypothesis Testing Procedure: Under the null hypothesis, the
approximated distribution of the deviance test statistic D~χ2 with df = n-p-1 = 316-3-1 = 312,
where n = # of observations, p = # of predicting variables.
25. C: L1, LASSO, forcing beta coefficients to zero; [5.7. Regularized Regression: Approaches]
Ridge regression does NOT perform variable selection because the variables remain in the
model. LASSO performs variable selection because coefficients will be forced to 0, which
removes the variable from the model. LASSO represents the L1 penalty.
26. D: 32; A model with p predictors has 2^p different model combinations.