ISYE 6501 FINAL EXAM– QUESTIONS AND ANSWERS |
VERIFIED AND WELL DETAILED ANSWERS PLUS RATIONALES |
GUARANTEED PASS | LATEST EXAM UPDATE | EXAM PREP |
STUDY GUIDE | PRACTICE TEST| DOWNLOAD INSTANT PDF
1. A manufacturing plant manager wants to forecast monthly electricity demand for the
upcoming year using historical monthly data spanning the last five years. The data clearly
shows a repeating seasonal pattern every 12 months alongside a gradual upward trend due
to increased production capacity. Which of the following modeling approaches is most
appropriate for this scenario?
A. Simple Exponential Smoothing
B. Holt-Winters Additive or Multiplicative Smoothing
C. Simple Linear Regression without seasonality terms
D. Moving Average with a window of 3 months
Holt-Winters exponential smoothing is explicitly designed to handle time series data that
exhibits both a trend component and a seasonal component. Simple exponential smoothing
only handles level without trend or seasonality, simple linear regression fails to capture
seasonal periodicity, and a 3-month moving average is too short-term and smooths out both
trend and seasonality.
2. An analyst is building a multiple linear regression model to predict housing prices. After
running the initial regression, a plot of the studentized residuals against the fitted values
reveals a distinct megaphone shape, where the spread of residuals increases as the
predicted values increase. What primary assumption of linear regression has been
violated?
A. Multicollinearity
B. Normality of residuals
C. Homoscedasticity
D. Independence of errors
The megaphone or funnel shape in residual plots indicates heteroscedasticity, meaning the
variance of the error terms is not constant across all levels of the predictor variables.
Multicollinearity relates to correlation among predictors, normality deals with the distribution
of residuals, and independence relates to autocorrelation.
,3. A healthcare analytics team is developing a logistic regression model to predict whether
a patient will be readmitted within 30 days of discharge. The model yields an area under
the ROC curve (AUC-ROC) of 0.82. What does this specific metric communicate regarding
the model's performance?
A. The model correctly classifies 82% of all patients as readmitted or not readmitted regardless
of the threshold chosen.
B. There is an 82% probability that a randomly chosen patient who was actually
readmitted will receive a higher predicted probability of readmission than a randomly
chosen patient who was not readmitted.
C. The false positive rate of the model is fixed at 18% across all decision thresholds.
D. The model's baseline accuracy compared to a naive classifier is exactly 82%.
The AUC-ROC measures the model's ability to discriminate between positive and negative
classes and has a direct probabilistic interpretation: the likelihood that a randomly selected
positive instance is ranked higher than a randomly selected negative instance. Option A is
incorrect because accuracy depends on the chosen classification threshold, and the other
choices misinterpret the AUC definition.
4. A supply chain optimization team is utilizing integer programming to decide which
warehouse locations to open out of ten possible sites. Each site has a fixed opening cost, and
customer demands must be met while minimizing total fixed and shipping costs. What type
of optimization problem is this?
A. Unconstrained continuous optimization
B. Mixed-integer linear programming (MILP)
C. Pure nonlinear network flow problem
D. Convex quadratic programming
Deciding whether to open a warehouse is a binary (0-1) decision variable, making it an integer
programming problem, while shipping quantities are continuous variables. Because the
objective function and constraints are linear, this constitutes a mixed-integer linear program.
Pure continuous, nonlinear, and quadratic models do not accurately fit this binary facility
location framework.
5. In K-means clustering, an analyst notices that one specific cluster contains 75% of all
data points while the other three clusters contain only a handful of points each. The
numerical scales of the original features varied widely before clustering. What step was
most likely omitted prior to running the algorithm?
,A. Computing the correlation matrix
B. Standardizing or scaling the features
C. Removing all categorical variables
D. Converting the distance metric from Euclidean to Manhattan
K-means relies heavily on Euclidean distance calculations. If features are on different scales
(e.g., income in tens of thousands versus age in decades), features with larger magnitudes
completely dominate the distance metric, skewing cluster assignments. Standardization
ensures all features contribute equally to the distance calculations.
6. A retail firm wants to segment its customer base based on purchasing frequency, average
basket size, and total annual spend. The analytics team decides to use hierarchical
clustering instead of K-means. Which of the following is a primary advantage of
hierarchical clustering over K-means for this application?
A. Hierarchical clustering requires the analyst to pre-specify the exact number of clusters before
execution.
B. Hierarchical clustering is computationally much faster on massive datasets containing
millions of rows.
C. Hierarchical clustering produces a dendrogram that allows the analyst to evaluate
multiple cluster solutions visually at different cut levels.
D. Hierarchical clustering is completely immune to the presence of outliers in the dataset.
A major strength of agglomerative hierarchical clustering is the dendrogram output, which
visualizes the merging process and lets decision-makers choose the optimal number of clusters
post-hoc. K-means requires pre-specifying k, is faster on massive datasets, and both methods
remain sensitive to outliers.
7. An e-commerce company is testing a new website checkout layout to see if it increases
average order value. They randomly assign incoming visitors to either the control group
(old layout) or the treatment group (new layout) and perform a two-sample t-test. The
resulting p-value is 0.03 at a significance level of alpha = 0.05. What is the correct statistical
conclusion?
A. Accept the null hypothesis and conclude the new layout has no effect on order value.
B. Reject the null hypothesis and conclude there is statistically significant evidence that the
new layout changes average order value.
, C. Prove that the new layout causes an increase in customer satisfaction scores across all
demographics.
D. Conclude that the probability the null hypothesis is true equals 3%.
Since the p-value (0.03) is less than the significance level (0.05), we reject the null hypothesis
of no difference. A p-value does not prove causation universally, nor does it represent the
probability that the null hypothesis is true; it represents the probability of observing the data
given that the null hypothesis is true.
8. A financial institution uses a Support Vector Machine (SVM) classifier with a radial
basis function (RBF) kernel to detect fraudulent transactions. During model tuning, the
team increases the regularization parameter C to an extremely large value. What is the
most likely consequence on model performance?
A. The model will underfit both training and test data due to excessive smoothing of the
decision boundary.
B. The model will overfit the training data by creating a highly complex, wiggly decision
boundary to minimize training errors.
C. The optimization solver will fail to converge because large C values violate kernel Mercer
conditions.
D. The decision boundary will automatically transform from non-linear to a strict linear
hyperplane.
In SVMs, the parameter C controls the penalty for misclassifications. A very large C tells the
model to heavily penalize misclassified training points, forcing the algorithm to create a
complex decision boundary that tightly hugs training data, leading to overfitting and poor
generalization on unseen data.
9. A regional logistics coordinator is analyzing delivery times across four different shipping
hubs. To determine if the mean delivery times differ significantly across the hubs, the
analyst performs a One-Way ANOVA. The resulting p-value is less than 0.001. What
analytical step should the analyst take next?
A. Immediately declare hub A as the slowest and hub C as the fastest without further testing.
B. Conduct post-hoc pairwise comparison tests (such as Tukey's HSD) to determine which
specific pairs of hubs have significantly different means.
C. Discard the ANOVA results because ANOVA cannot be used when comparing more than
two groups.
D. Run a simple linear regression using hub names as continuous numeric variables.
VERIFIED AND WELL DETAILED ANSWERS PLUS RATIONALES |
GUARANTEED PASS | LATEST EXAM UPDATE | EXAM PREP |
STUDY GUIDE | PRACTICE TEST| DOWNLOAD INSTANT PDF
1. A manufacturing plant manager wants to forecast monthly electricity demand for the
upcoming year using historical monthly data spanning the last five years. The data clearly
shows a repeating seasonal pattern every 12 months alongside a gradual upward trend due
to increased production capacity. Which of the following modeling approaches is most
appropriate for this scenario?
A. Simple Exponential Smoothing
B. Holt-Winters Additive or Multiplicative Smoothing
C. Simple Linear Regression without seasonality terms
D. Moving Average with a window of 3 months
Holt-Winters exponential smoothing is explicitly designed to handle time series data that
exhibits both a trend component and a seasonal component. Simple exponential smoothing
only handles level without trend or seasonality, simple linear regression fails to capture
seasonal periodicity, and a 3-month moving average is too short-term and smooths out both
trend and seasonality.
2. An analyst is building a multiple linear regression model to predict housing prices. After
running the initial regression, a plot of the studentized residuals against the fitted values
reveals a distinct megaphone shape, where the spread of residuals increases as the
predicted values increase. What primary assumption of linear regression has been
violated?
A. Multicollinearity
B. Normality of residuals
C. Homoscedasticity
D. Independence of errors
The megaphone or funnel shape in residual plots indicates heteroscedasticity, meaning the
variance of the error terms is not constant across all levels of the predictor variables.
Multicollinearity relates to correlation among predictors, normality deals with the distribution
of residuals, and independence relates to autocorrelation.
,3. A healthcare analytics team is developing a logistic regression model to predict whether
a patient will be readmitted within 30 days of discharge. The model yields an area under
the ROC curve (AUC-ROC) of 0.82. What does this specific metric communicate regarding
the model's performance?
A. The model correctly classifies 82% of all patients as readmitted or not readmitted regardless
of the threshold chosen.
B. There is an 82% probability that a randomly chosen patient who was actually
readmitted will receive a higher predicted probability of readmission than a randomly
chosen patient who was not readmitted.
C. The false positive rate of the model is fixed at 18% across all decision thresholds.
D. The model's baseline accuracy compared to a naive classifier is exactly 82%.
The AUC-ROC measures the model's ability to discriminate between positive and negative
classes and has a direct probabilistic interpretation: the likelihood that a randomly selected
positive instance is ranked higher than a randomly selected negative instance. Option A is
incorrect because accuracy depends on the chosen classification threshold, and the other
choices misinterpret the AUC definition.
4. A supply chain optimization team is utilizing integer programming to decide which
warehouse locations to open out of ten possible sites. Each site has a fixed opening cost, and
customer demands must be met while minimizing total fixed and shipping costs. What type
of optimization problem is this?
A. Unconstrained continuous optimization
B. Mixed-integer linear programming (MILP)
C. Pure nonlinear network flow problem
D. Convex quadratic programming
Deciding whether to open a warehouse is a binary (0-1) decision variable, making it an integer
programming problem, while shipping quantities are continuous variables. Because the
objective function and constraints are linear, this constitutes a mixed-integer linear program.
Pure continuous, nonlinear, and quadratic models do not accurately fit this binary facility
location framework.
5. In K-means clustering, an analyst notices that one specific cluster contains 75% of all
data points while the other three clusters contain only a handful of points each. The
numerical scales of the original features varied widely before clustering. What step was
most likely omitted prior to running the algorithm?
,A. Computing the correlation matrix
B. Standardizing or scaling the features
C. Removing all categorical variables
D. Converting the distance metric from Euclidean to Manhattan
K-means relies heavily on Euclidean distance calculations. If features are on different scales
(e.g., income in tens of thousands versus age in decades), features with larger magnitudes
completely dominate the distance metric, skewing cluster assignments. Standardization
ensures all features contribute equally to the distance calculations.
6. A retail firm wants to segment its customer base based on purchasing frequency, average
basket size, and total annual spend. The analytics team decides to use hierarchical
clustering instead of K-means. Which of the following is a primary advantage of
hierarchical clustering over K-means for this application?
A. Hierarchical clustering requires the analyst to pre-specify the exact number of clusters before
execution.
B. Hierarchical clustering is computationally much faster on massive datasets containing
millions of rows.
C. Hierarchical clustering produces a dendrogram that allows the analyst to evaluate
multiple cluster solutions visually at different cut levels.
D. Hierarchical clustering is completely immune to the presence of outliers in the dataset.
A major strength of agglomerative hierarchical clustering is the dendrogram output, which
visualizes the merging process and lets decision-makers choose the optimal number of clusters
post-hoc. K-means requires pre-specifying k, is faster on massive datasets, and both methods
remain sensitive to outliers.
7. An e-commerce company is testing a new website checkout layout to see if it increases
average order value. They randomly assign incoming visitors to either the control group
(old layout) or the treatment group (new layout) and perform a two-sample t-test. The
resulting p-value is 0.03 at a significance level of alpha = 0.05. What is the correct statistical
conclusion?
A. Accept the null hypothesis and conclude the new layout has no effect on order value.
B. Reject the null hypothesis and conclude there is statistically significant evidence that the
new layout changes average order value.
, C. Prove that the new layout causes an increase in customer satisfaction scores across all
demographics.
D. Conclude that the probability the null hypothesis is true equals 3%.
Since the p-value (0.03) is less than the significance level (0.05), we reject the null hypothesis
of no difference. A p-value does not prove causation universally, nor does it represent the
probability that the null hypothesis is true; it represents the probability of observing the data
given that the null hypothesis is true.
8. A financial institution uses a Support Vector Machine (SVM) classifier with a radial
basis function (RBF) kernel to detect fraudulent transactions. During model tuning, the
team increases the regularization parameter C to an extremely large value. What is the
most likely consequence on model performance?
A. The model will underfit both training and test data due to excessive smoothing of the
decision boundary.
B. The model will overfit the training data by creating a highly complex, wiggly decision
boundary to minimize training errors.
C. The optimization solver will fail to converge because large C values violate kernel Mercer
conditions.
D. The decision boundary will automatically transform from non-linear to a strict linear
hyperplane.
In SVMs, the parameter C controls the penalty for misclassifications. A very large C tells the
model to heavily penalize misclassified training points, forcing the algorithm to create a
complex decision boundary that tightly hugs training data, leading to overfitting and poor
generalization on unseen data.
9. A regional logistics coordinator is analyzing delivery times across four different shipping
hubs. To determine if the mean delivery times differ significantly across the hubs, the
analyst performs a One-Way ANOVA. The resulting p-value is less than 0.001. What
analytical step should the analyst take next?
A. Immediately declare hub A as the slowest and hub C as the fastest without further testing.
B. Conduct post-hoc pairwise comparison tests (such as Tukey's HSD) to determine which
specific pairs of hubs have significantly different means.
C. Discard the ANOVA results because ANOVA cannot be used when comparing more than
two groups.
D. Run a simple linear regression using hub names as continuous numeric variables.
