1|Page
ISYE 6501 Final Exam | Latest Update | Comprehensive Practice
Questions & Answers | Introduction to Analytics Modeling Study
Guide & Exam Preparation PDF
ISYE 6501 Final Exam | Comprehensive Practice Questions
Exam Coverage Summary
This examination covers the fundamental concepts of Introduction to Analytics Modeling including optimization
techniques (linear programming, integer programming, network models), supervised learning methods (regression,
classification, tree-based models, support vector machines), unsupervised learning approaches (clustering, principal
component analysis), time series analysis (ARIMA, exponential smoothing, seasonality), simulation modeling
(discrete-event simulation, queuing theory), validation techniques (cross-validation, training/testing splits),
probability distributions (Bernoulli, binomial, Poisson, exponential, normal), distance metrics (Euclidean,
Manhattan, infinity norm), model selection criteria (AIC, BIC, adjusted R-squared), regularization methods (ridge,
lasso, elastic net), experimental design (factorial designs, blocking), change detection methods (CUSUM), game
theory concepts (prisoner's dilemma, zero-sum games), dynamic programming, and analytics taxonomy (descriptive,
predictive, prescriptive). The exam emphasizes practical application of these concepts through scenario-based
questions that test understanding of when and how to apply various analytical techniques in real-world business and
engineering contexts.
1. A logistics manager needs to assign five delivery trucks to five different routes, with each truck assigned to
exactly one route, minimizing total fuel costs. Which optimization model should they use?
A) Shortest path problem
B) Maximum flow problem
C) Assignment problem
D) Diet problem
C) Assignment problem
Assignment problems involve matching items from two sets one-to-one to minimize or maximize total cost, making
them ideal for truck-to-route allocation scenarios.
2. An analyst evaluating a credit scoring model finds that 85% of actual defaulters were correctly identified. What
metric does this represent?
,2|Page
A) Specificity
B) Precision
C) Sensitivity
D) False positive rate
C) Sensitivity
Sensitivity, also called recall or true positive rate, measures the proportion of actual positives correctly identified by
the model, calculated as TP divided by TP plus FN.
3. When comparing two regression models, Model A has an R-squared of 0.85 with 8 predictors and Model B has an
R-squared of 0.84 with 3 predictors. Which metric would properly penalize Model A for its additional complexity?
A) R-squared
B) Adjusted R-squared
C) Correlation coefficient
D) P-value
B) Adjusted R-squared
Adjusted R-squared incorporates a penalty for each additional predictor variable, making it useful for comparing
models with different numbers of variables and encouraging simpler models.
4. A manufacturing plant wants to decide whether to build a new facility based on uncertain future demand. They
consider both the best and worst possible outcomes for each decision. What type of optimization approach is being
used?
A) Stochastic optimization
B) Robust optimization
C) Convex optimization
D) Dynamic programming
B) Robust optimization
Robust optimization focuses on finding solutions that perform acceptably under worst-case scenarios, making it
appropriate when decision-makers want to protect against the most unfavorable outcomes.
5. In a time series with monthly sales data showing consistent increases and recurring yearly patterns, which three
components should be modeled?
,3|Page
A) Trend, seasonality, and error
B) Mean, variance, and autocorrelation
C) Cycle, noise, and stationarity
D) Level, slope, and intercept
A) Trend, seasonality, and error
Time series data typically contains trend components, seasonal patterns that repeat at fixed intervals, and random
error or noise that cannot be explained by the systematic components.
6. A customer service center receives calls according to a Poisson process with an average rate of 10 calls per hour.
What is the distribution of the time between consecutive calls?
A) Normal distribution
B) Exponential distribution
C) Uniform distribution
D) Binomial distribution
B) Exponential distribution
If arrivals follow a Poisson process, the interarrival times follow an exponential distribution with the same rate
parameter, demonstrating the memoryless property of the exponential distribution.
7. In k-means clustering, what determines the number of clusters to use?
A) The dimensionality of the data
B) The total number of data points
C) Analysis of the elbow diagram
D) The maximum distance between points
C) Analysis of the elbow diagram
The elbow diagram plots within-cluster sum of squares against the number of clusters, and the optimal k is typically
chosen at the point where the rate of improvement decreases significantly.
8. A binary classification model has the following confusion matrix: TP=80, FP=20, TN=170, FN=30. What is the
accuracy of this model?
A) 0.80
B) 0.83
, 4|Page
C) 0.85
D) 0.88
B) 0.83
Accuracy equals TP plus TN divided by the total number of observations, which is 80 plus 170 equals 250, divided
by 80 plus 20 plus 170 plus 30 equals 300, resulting in 0.833.
9. Which property distinguishes the exponential distribution from most other continuous probability distributions?
A) It is always symmetric
B) It has no memory
C) It has a finite support
D) It is discrete
B) It has no memory
The exponential distribution exhibits the memoryless property, meaning the probability of an event occurring in the
future is independent of how much time has already passed without an event.
10. A data scientist builds a model with 20 predictors and only 50 data points, achieving perfect classification on
training data but poor performance on test data. What problem has occurred?
A) Underfitting
B) Overfitting
C) Multicollinearity
D) Heteroscedasticity
B) Overfitting
Overfitting occurs when a model captures random noise rather than real patterns, often happening when the number
of parameters exceeds the number of data points, leading to poor generalization to new data.
11. In a support vector machine, what are the data points that are closest to the decision boundary called?
A) Centroid points
B) Support vectors
C) Kernel points
D) Margin points
ISYE 6501 Final Exam | Latest Update | Comprehensive Practice
Questions & Answers | Introduction to Analytics Modeling Study
Guide & Exam Preparation PDF
ISYE 6501 Final Exam | Comprehensive Practice Questions
Exam Coverage Summary
This examination covers the fundamental concepts of Introduction to Analytics Modeling including optimization
techniques (linear programming, integer programming, network models), supervised learning methods (regression,
classification, tree-based models, support vector machines), unsupervised learning approaches (clustering, principal
component analysis), time series analysis (ARIMA, exponential smoothing, seasonality), simulation modeling
(discrete-event simulation, queuing theory), validation techniques (cross-validation, training/testing splits),
probability distributions (Bernoulli, binomial, Poisson, exponential, normal), distance metrics (Euclidean,
Manhattan, infinity norm), model selection criteria (AIC, BIC, adjusted R-squared), regularization methods (ridge,
lasso, elastic net), experimental design (factorial designs, blocking), change detection methods (CUSUM), game
theory concepts (prisoner's dilemma, zero-sum games), dynamic programming, and analytics taxonomy (descriptive,
predictive, prescriptive). The exam emphasizes practical application of these concepts through scenario-based
questions that test understanding of when and how to apply various analytical techniques in real-world business and
engineering contexts.
1. A logistics manager needs to assign five delivery trucks to five different routes, with each truck assigned to
exactly one route, minimizing total fuel costs. Which optimization model should they use?
A) Shortest path problem
B) Maximum flow problem
C) Assignment problem
D) Diet problem
C) Assignment problem
Assignment problems involve matching items from two sets one-to-one to minimize or maximize total cost, making
them ideal for truck-to-route allocation scenarios.
2. An analyst evaluating a credit scoring model finds that 85% of actual defaulters were correctly identified. What
metric does this represent?
,2|Page
A) Specificity
B) Precision
C) Sensitivity
D) False positive rate
C) Sensitivity
Sensitivity, also called recall or true positive rate, measures the proportion of actual positives correctly identified by
the model, calculated as TP divided by TP plus FN.
3. When comparing two regression models, Model A has an R-squared of 0.85 with 8 predictors and Model B has an
R-squared of 0.84 with 3 predictors. Which metric would properly penalize Model A for its additional complexity?
A) R-squared
B) Adjusted R-squared
C) Correlation coefficient
D) P-value
B) Adjusted R-squared
Adjusted R-squared incorporates a penalty for each additional predictor variable, making it useful for comparing
models with different numbers of variables and encouraging simpler models.
4. A manufacturing plant wants to decide whether to build a new facility based on uncertain future demand. They
consider both the best and worst possible outcomes for each decision. What type of optimization approach is being
used?
A) Stochastic optimization
B) Robust optimization
C) Convex optimization
D) Dynamic programming
B) Robust optimization
Robust optimization focuses on finding solutions that perform acceptably under worst-case scenarios, making it
appropriate when decision-makers want to protect against the most unfavorable outcomes.
5. In a time series with monthly sales data showing consistent increases and recurring yearly patterns, which three
components should be modeled?
,3|Page
A) Trend, seasonality, and error
B) Mean, variance, and autocorrelation
C) Cycle, noise, and stationarity
D) Level, slope, and intercept
A) Trend, seasonality, and error
Time series data typically contains trend components, seasonal patterns that repeat at fixed intervals, and random
error or noise that cannot be explained by the systematic components.
6. A customer service center receives calls according to a Poisson process with an average rate of 10 calls per hour.
What is the distribution of the time between consecutive calls?
A) Normal distribution
B) Exponential distribution
C) Uniform distribution
D) Binomial distribution
B) Exponential distribution
If arrivals follow a Poisson process, the interarrival times follow an exponential distribution with the same rate
parameter, demonstrating the memoryless property of the exponential distribution.
7. In k-means clustering, what determines the number of clusters to use?
A) The dimensionality of the data
B) The total number of data points
C) Analysis of the elbow diagram
D) The maximum distance between points
C) Analysis of the elbow diagram
The elbow diagram plots within-cluster sum of squares against the number of clusters, and the optimal k is typically
chosen at the point where the rate of improvement decreases significantly.
8. A binary classification model has the following confusion matrix: TP=80, FP=20, TN=170, FN=30. What is the
accuracy of this model?
A) 0.80
B) 0.83
, 4|Page
C) 0.85
D) 0.88
B) 0.83
Accuracy equals TP plus TN divided by the total number of observations, which is 80 plus 170 equals 250, divided
by 80 plus 20 plus 170 plus 30 equals 300, resulting in 0.833.
9. Which property distinguishes the exponential distribution from most other continuous probability distributions?
A) It is always symmetric
B) It has no memory
C) It has a finite support
D) It is discrete
B) It has no memory
The exponential distribution exhibits the memoryless property, meaning the probability of an event occurring in the
future is independent of how much time has already passed without an event.
10. A data scientist builds a model with 20 predictors and only 50 data points, achieving perfect classification on
training data but poor performance on test data. What problem has occurred?
A) Underfitting
B) Overfitting
C) Multicollinearity
D) Heteroscedasticity
B) Overfitting
Overfitting occurs when a model captures random noise rather than real patterns, often happening when the number
of parameters exceeds the number of data points, leading to poor generalization to new data.
11. In a support vector machine, what are the data points that are closest to the decision boundary called?
A) Centroid points
B) Support vectors
C) Kernel points
D) Margin points
