ISYE 6501 ANALYTICS MODELING
COMPREHENSIVE FINAL
EXAMINATION 2026 QUESTIONS AND
ANSWERS
1. In Support Vector Machines (SVM), what is the primary purpose of the ‘soft margin’
parameter (often denoted as C)?
A. To control the trade-off between maximizing the margin and minimizing classification
errors.
B. To ensure that the hyperplane perfectly separates all training data points.
C. To transform non-linearly separable data into a higher-dimensional space.
D. To eliminate the need for kernel functions in complex datasets.
Answer: A
Conceptual Explanation: The parameter C regulates the balance between maximizing the
width of the margin and minimizing the number of misclassifications in the training set.
2. Which of the following scenarios is most indicative of overfitting in a regression model?
A. High R-squared on the training set and low R-squared on the test set.
B. Low R-squared on the training set and high R-squared on the test set.
,C. High R-squared on the training set and high R-squared on the test set.
D. Low R-squared on both the training and test sets.
Answer: A
Conceptual Explanation: Overfitting occurs when a model captures noise as if it were a
signal, leading to excellent performance on training data but poor generalization to unseen
data.
3. In the context of K-means clustering, why is it often necessary to run the algorithm
multiple times with different initializations?
A. To ensure that every data point belongs to at least two different clusters.
B. Because K-means is guaranteed to find a global optimum only after 10 iterations.
C. To determine the optimal value of K through the Elbow method.
D. To overcome the algorithm’s sensitivity to the initial placement of centroids, which can
lead to local optima.
Answer: D
Conceptual Explanation: K-means is a heuristic algorithm that converges to local optima.
Different starting points can yield different final clusterings, so multiple runs help find a
better solution.
, 4. When using the Holt-Winters method for time series forecasting, which component is
added when moving from Double to Triple Exponential Smoothing?
A. Seasonality
B. Level
C. Trend
D. Autoregression
Answer: A
Conceptual Explanation: Single smoothing handles level, Double handles level and trend,
and Triple (Holt-Winters) handles level, trend, and seasonality.
5. Which regularization technique can be used for feature selection because it forces some
coefficients to be exactly zero?
A. Ridge Regression (L2)
B. Polynomial Regression
C. Ordinary Least Squares
D. LASSO Regression (L1)
Answer: D
Conceptual Explanation: LASSO (Least Absolute Shrinkage and Selection Operator) uses
the L1 penalty, which has the mathematical property of shrinking less important
coefficients to zero.
COMPREHENSIVE FINAL
EXAMINATION 2026 QUESTIONS AND
ANSWERS
1. In Support Vector Machines (SVM), what is the primary purpose of the ‘soft margin’
parameter (often denoted as C)?
A. To control the trade-off between maximizing the margin and minimizing classification
errors.
B. To ensure that the hyperplane perfectly separates all training data points.
C. To transform non-linearly separable data into a higher-dimensional space.
D. To eliminate the need for kernel functions in complex datasets.
Answer: A
Conceptual Explanation: The parameter C regulates the balance between maximizing the
width of the margin and minimizing the number of misclassifications in the training set.
2. Which of the following scenarios is most indicative of overfitting in a regression model?
A. High R-squared on the training set and low R-squared on the test set.
B. Low R-squared on the training set and high R-squared on the test set.
,C. High R-squared on the training set and high R-squared on the test set.
D. Low R-squared on both the training and test sets.
Answer: A
Conceptual Explanation: Overfitting occurs when a model captures noise as if it were a
signal, leading to excellent performance on training data but poor generalization to unseen
data.
3. In the context of K-means clustering, why is it often necessary to run the algorithm
multiple times with different initializations?
A. To ensure that every data point belongs to at least two different clusters.
B. Because K-means is guaranteed to find a global optimum only after 10 iterations.
C. To determine the optimal value of K through the Elbow method.
D. To overcome the algorithm’s sensitivity to the initial placement of centroids, which can
lead to local optima.
Answer: D
Conceptual Explanation: K-means is a heuristic algorithm that converges to local optima.
Different starting points can yield different final clusterings, so multiple runs help find a
better solution.
, 4. When using the Holt-Winters method for time series forecasting, which component is
added when moving from Double to Triple Exponential Smoothing?
A. Seasonality
B. Level
C. Trend
D. Autoregression
Answer: A
Conceptual Explanation: Single smoothing handles level, Double handles level and trend,
and Triple (Holt-Winters) handles level, trend, and seasonality.
5. Which regularization technique can be used for feature selection because it forces some
coefficients to be exactly zero?
A. Ridge Regression (L2)
B. Polynomial Regression
C. Ordinary Least Squares
D. LASSO Regression (L1)
Answer: D
Conceptual Explanation: LASSO (Least Absolute Shrinkage and Selection Operator) uses
the L1 penalty, which has the mathematical property of shrinking less important
coefficients to zero.
