ISYE 6501 Final Quiz | Questions and Answers | 2025 Update
| 100% Correct.
1. Which of the following is an example of a supervised learning problem?
A) Grouping customers into segments based on purchasing behavior
B) Predicting house prices based on square footage and number of bedrooms
C) Reducing the dimensionality of a dataset while preserving variance
D) Finding association rules in market basket data
Answer: B
Explanation: Supervised learning uses labeled data to predict an outcome. Predicting house prices (a
continuous outcome) from features is supervised regression. Clustering (A) and association rules (D) are
unsupervised. PCA (C) is dimensionality reduction.
2. In logistic regression, the link function used is typically the:
A) Identity function
B) Log function
C) Logit function
D) Exponential function
Answer: C
Explanation: Logistic regression uses the logit link function: log(p/(1-p)) = β₀ + β₁X. This maps
probabilities (0 to 1) to the entire real line.
3. Which of the following is NOT an appropriate evaluation metric for a binary classification model on an
imbalanced dataset where the minority class is the most important?
A) Accuracy
B) Precision
C) Recall
D) F1-score
Answer: A
,Explanation: Accuracy can be misleading on imbalanced data (e.g., 95% accuracy by predicting all
majority class). Precision, recall, and F1 are better for imbalanced problems.
4. In k-nearest neighbors (KNN), as k increases, the decision boundary becomes:
A) More complex and jagged
B) Simpler and smoother
C) Unchanged
D) Non-existent
Answer: B
Explanation: For KNN, a small k leads to a complex, overfitted decision boundary. As k increases, the
model averages over more neighbors, making the boundary smoother and less complex.
5. Which of the following is an example of an unsupervised learning task?
A) Predicting whether an email is spam or not
B) Estimating house prices
C) Grouping news articles into topics
D) Forecasting stock prices
Answer: C
Explanation: Topic modeling (grouping articles into topics without labels) is unsupervised. Spam
detection, price prediction, and stock forecasting are supervised.
6. In the context of support vector machines (SVM), the "kernel trick" allows:
A) Faster training time
B) Handling of non-linear decision boundaries without explicit transformation
C) Reduction in the number of support vectors
D) Automatic feature selection
Answer: B
Explanation: The kernel trick implicitly maps data to a higher-dimensional space to find linear
separability, avoiding explicit computation of the transformation.
,7. Which clustering algorithm requires the number of clusters (k) to be specified in advance?
A) Hierarchical clustering
B) DBSCAN
C) K-means
D) Gaussian mixture models (with known components)
Answer: C
Explanation: K-means requires the user to specify k. Hierarchical clustering does not require k upfront
(can be chosen from dendrogram). DBSCAN uses density parameters.
8. The elbow method is used to:
A) Select the number of clusters in k-means
B) Choose the optimal k in KNN
C) Determine the regularization parameter in LASSO
D) Find the best kernel for SVM
Answer: A
Explanation: The elbow method plots within-cluster sum of squares (WCSS) vs. number of clusters; the
"elbow" suggests an appropriate k.
9. The area under the ROC curve (AUC) measures:
A) The accuracy of the model at a single threshold
B) The model's ability to distinguish between positive and negative classes across all thresholds
C) The precision of the model
D) The recall of the model
Answer: B
Explanation: AUC represents the probability that a randomly chosen positive instance is ranked higher
than a randomly chosen negative instance.
10. In LASSO regression (L1 regularization), the penalty term is proportional to the:
A) Sum of squared coefficients
B) Sum of absolute values of coefficients
, C) Sum of coefficients
D) Square root of sum of coefficients
Answer: B
Explanation: L1 regularization adds penalty = λ Σ|βⱼ|. Ridge (L2) uses sum of squared coefficients Σβⱼ².
11. Principal Component Analysis (PCA) is primarily used for:
A) Classification
B) Clustering
C) Dimensionality reduction while preserving variance
D) Time series forecasting
Answer: C
Explanation: PCA finds orthogonal linear combinations of features that maximize variance, reducing
dimensionality with minimal information loss.
12. In a decision tree, the Gini impurity at a node with 80% class A and 20% class B is:
A) 0.16
B) 0.32
C) 0.48
D) 0.64
Answer: B
Explanation: Gini = 1 - (0.8² + 0.2²) = 1 - (0.64 + 0.04) = 1 - 0.68 = 0.32.
13. In exponential smoothing, a larger smoothing parameter (α close to 1) means:
A) More weight is given to past observations
B) More weight is given to the most recent observation
C) The forecast is flat
D) Seasonality is ignored
Answer: B
| 100% Correct.
1. Which of the following is an example of a supervised learning problem?
A) Grouping customers into segments based on purchasing behavior
B) Predicting house prices based on square footage and number of bedrooms
C) Reducing the dimensionality of a dataset while preserving variance
D) Finding association rules in market basket data
Answer: B
Explanation: Supervised learning uses labeled data to predict an outcome. Predicting house prices (a
continuous outcome) from features is supervised regression. Clustering (A) and association rules (D) are
unsupervised. PCA (C) is dimensionality reduction.
2. In logistic regression, the link function used is typically the:
A) Identity function
B) Log function
C) Logit function
D) Exponential function
Answer: C
Explanation: Logistic regression uses the logit link function: log(p/(1-p)) = β₀ + β₁X. This maps
probabilities (0 to 1) to the entire real line.
3. Which of the following is NOT an appropriate evaluation metric for a binary classification model on an
imbalanced dataset where the minority class is the most important?
A) Accuracy
B) Precision
C) Recall
D) F1-score
Answer: A
,Explanation: Accuracy can be misleading on imbalanced data (e.g., 95% accuracy by predicting all
majority class). Precision, recall, and F1 are better for imbalanced problems.
4. In k-nearest neighbors (KNN), as k increases, the decision boundary becomes:
A) More complex and jagged
B) Simpler and smoother
C) Unchanged
D) Non-existent
Answer: B
Explanation: For KNN, a small k leads to a complex, overfitted decision boundary. As k increases, the
model averages over more neighbors, making the boundary smoother and less complex.
5. Which of the following is an example of an unsupervised learning task?
A) Predicting whether an email is spam or not
B) Estimating house prices
C) Grouping news articles into topics
D) Forecasting stock prices
Answer: C
Explanation: Topic modeling (grouping articles into topics without labels) is unsupervised. Spam
detection, price prediction, and stock forecasting are supervised.
6. In the context of support vector machines (SVM), the "kernel trick" allows:
A) Faster training time
B) Handling of non-linear decision boundaries without explicit transformation
C) Reduction in the number of support vectors
D) Automatic feature selection
Answer: B
Explanation: The kernel trick implicitly maps data to a higher-dimensional space to find linear
separability, avoiding explicit computation of the transformation.
,7. Which clustering algorithm requires the number of clusters (k) to be specified in advance?
A) Hierarchical clustering
B) DBSCAN
C) K-means
D) Gaussian mixture models (with known components)
Answer: C
Explanation: K-means requires the user to specify k. Hierarchical clustering does not require k upfront
(can be chosen from dendrogram). DBSCAN uses density parameters.
8. The elbow method is used to:
A) Select the number of clusters in k-means
B) Choose the optimal k in KNN
C) Determine the regularization parameter in LASSO
D) Find the best kernel for SVM
Answer: A
Explanation: The elbow method plots within-cluster sum of squares (WCSS) vs. number of clusters; the
"elbow" suggests an appropriate k.
9. The area under the ROC curve (AUC) measures:
A) The accuracy of the model at a single threshold
B) The model's ability to distinguish between positive and negative classes across all thresholds
C) The precision of the model
D) The recall of the model
Answer: B
Explanation: AUC represents the probability that a randomly chosen positive instance is ranked higher
than a randomly chosen negative instance.
10. In LASSO regression (L1 regularization), the penalty term is proportional to the:
A) Sum of squared coefficients
B) Sum of absolute values of coefficients
, C) Sum of coefficients
D) Square root of sum of coefficients
Answer: B
Explanation: L1 regularization adds penalty = λ Σ|βⱼ|. Ridge (L2) uses sum of squared coefficients Σβⱼ².
11. Principal Component Analysis (PCA) is primarily used for:
A) Classification
B) Clustering
C) Dimensionality reduction while preserving variance
D) Time series forecasting
Answer: C
Explanation: PCA finds orthogonal linear combinations of features that maximize variance, reducing
dimensionality with minimal information loss.
12. In a decision tree, the Gini impurity at a node with 80% class A and 20% class B is:
A) 0.16
B) 0.32
C) 0.48
D) 0.64
Answer: B
Explanation: Gini = 1 - (0.8² + 0.2²) = 1 - (0.64 + 0.04) = 1 - 0.68 = 0.32.
13. In exponential smoothing, a larger smoothing parameter (α close to 1) means:
A) More weight is given to past observations
B) More weight is given to the most recent observation
C) The forecast is flat
D) Seasonality is ignored
Answer: B
