ISYE 6501 FINAL EXAM - ADVANCED
MODELING ANALYTICS QUESTIONS
AND ANSWERS
1. When using K-fold cross-validation, how does increasing the number of folds (K) typically
impact the bias and variance of the model performance estimate?
A. Bias increases while variance decreases.
B. Both bias and variance decrease.
C. Bias decreases while variance increases.
D. Both bias and variance increase.
Answer: C
Conceptual Explanation: As K increases, the size of the training set in each fold
approaches the size of the full dataset, which reduces bias. However, since the training sets
become more similar (highly correlated), the variance of the performance estimate across
the folds increases.
2. In Support Vector Machines (SVM), what is the primary role of the C parameter in the soft-
margin formulation?
A. It defines the degree of the polynomial kernel.
,B. It controls the trade-off between maximizing the margin and minimizing classification
errors.
C. It determines the number of support vectors to be ignored.
D. It scales the input features to a range between 0 and 1.
Answer: B
Conceptual Explanation: The C parameter is a regularization parameter that determines
the weight given to misclassifications. A small C allows more misclassifications for a wider
margin, while a large C penalizes misclassifications heavily, leading to a narrower margin.
3. Which of the following scenarios best justifies the use of Lasso (L1) regression over Ridge
(L2) regression?
A. When most features are expected to have a small but non-zero influence on the target.
B. When you have a very large number of features and suspect only a few are truly
relevant.
C. When there is significant multicollinearity and you want to keep all variables.
D. When the target variable follows a Poisson distribution.
Answer: B
Conceptual Explanation: Lasso regression provides automatic feature selection by
shrinking the coefficients of less important variables exactly to zero, which is useful when
sparse solutions are preferred.
, 4. In the context of the CUSUM (Cumulative Sum) method for change detection, what is the
effect of increasing the drift parameter ‘k’?
A. The detector becomes more sensitive to small shifts in the mean.
B. The calculation of the cumulative sum becomes non-linear.
C. The threshold ‘H’ is automatically lowered.
D. The detector becomes less sensitive to small shifts, reducing false alarms.
Answer: D
Conceptual Explanation: The parameter ‘k’ represents the allowable slack or drift.
Increasing ‘k’ means the process mean must shift by a larger amount before the CUSUM
starts accumulating, making it less sensitive to minor fluctuations.
5. Which component of the Holt-Winters exponential smoothing method is specifically
designed to handle multiplicative seasonality?
A. Level (alpha)
B. Trend (beta)
C. The Autoregressive term (phi)
D. The Seasonality (gamma) factor applied as a ratio rather than an addition.
Answer: D
MODELING ANALYTICS QUESTIONS
AND ANSWERS
1. When using K-fold cross-validation, how does increasing the number of folds (K) typically
impact the bias and variance of the model performance estimate?
A. Bias increases while variance decreases.
B. Both bias and variance decrease.
C. Bias decreases while variance increases.
D. Both bias and variance increase.
Answer: C
Conceptual Explanation: As K increases, the size of the training set in each fold
approaches the size of the full dataset, which reduces bias. However, since the training sets
become more similar (highly correlated), the variance of the performance estimate across
the folds increases.
2. In Support Vector Machines (SVM), what is the primary role of the C parameter in the soft-
margin formulation?
A. It defines the degree of the polynomial kernel.
,B. It controls the trade-off between maximizing the margin and minimizing classification
errors.
C. It determines the number of support vectors to be ignored.
D. It scales the input features to a range between 0 and 1.
Answer: B
Conceptual Explanation: The C parameter is a regularization parameter that determines
the weight given to misclassifications. A small C allows more misclassifications for a wider
margin, while a large C penalizes misclassifications heavily, leading to a narrower margin.
3. Which of the following scenarios best justifies the use of Lasso (L1) regression over Ridge
(L2) regression?
A. When most features are expected to have a small but non-zero influence on the target.
B. When you have a very large number of features and suspect only a few are truly
relevant.
C. When there is significant multicollinearity and you want to keep all variables.
D. When the target variable follows a Poisson distribution.
Answer: B
Conceptual Explanation: Lasso regression provides automatic feature selection by
shrinking the coefficients of less important variables exactly to zero, which is useful when
sparse solutions are preferred.
, 4. In the context of the CUSUM (Cumulative Sum) method for change detection, what is the
effect of increasing the drift parameter ‘k’?
A. The detector becomes more sensitive to small shifts in the mean.
B. The calculation of the cumulative sum becomes non-linear.
C. The threshold ‘H’ is automatically lowered.
D. The detector becomes less sensitive to small shifts, reducing false alarms.
Answer: D
Conceptual Explanation: The parameter ‘k’ represents the allowable slack or drift.
Increasing ‘k’ means the process mean must shift by a larger amount before the CUSUM
starts accumulating, making it less sensitive to minor fluctuations.
5. Which component of the Holt-Winters exponential smoothing method is specifically
designed to handle multiplicative seasonality?
A. Level (alpha)
B. Trend (beta)
C. The Autoregressive term (phi)
D. The Seasonality (gamma) factor applied as a ratio rather than an addition.
Answer: D
