ISYE 6501 Final Exam Review | Advanced Analytics Modeling
Practice Examination
Comprehensive Mastery Assessment - Second Edition
Exam Title: ISYE 6501 Comprehensive Final Examination: Advanced Analytics Modeling Practice
Assessment Covering Supervised Learning, Unsupervised Learning, Time Series Analysis, Optimization
Methods, Simulation Techniques, Experimental Design, and Model Validation for Graduate-Level
Analytics Practitioners
SECTION 1: FUNDAMENTAL CONCEPTS AND ANALYTICS
FRAMEWORK (Questions 1-20)
Question 1:
In the analytics framework, which type of question addresses "What actions should be taken to achieve
desired outcomes?"
A) Descriptive analytics
B) Predictive analytics
C) Prescriptive analytics
D) Diagnostic analytics
Correct Answer: C
Rationale: Prescriptive analytics focuses on recommending actions and decisions to achieve specific
goals. It answers "what should we do?" by using optimization, simulation, and decision analysis to
guide decision-making. Descriptive analytics describes what happened, diagnostic analytics explains
why it happened, and predictive analytics forecasts what will happen.
Question 2:
Which of the following best represents a descriptive analytics question?
A) How many units should we produce next month?
B) Which customers are most likely to respond to our marketing campaign?
C) What were our total sales by region last quarter?
D) Why did sales decline in the Northeast region?
Correct Answer: C
,Rationale: Descriptive analytics summarizes historical data to understand what has happened. Asking
about total sales by region in the past quarter is a descriptive question. Predictive questions forecast
future outcomes, prescriptive questions recommend actions, and diagnostic questions explore root
causes.
Question 3:
A retail company wants to predict next month's sales based on historical data. This is an example of:
A) Descriptive analytics
B) Predictive analytics
C) Prescriptive analytics
D) Diagnostic analytics
Correct Answer: B
Rationale: Predictive analytics uses historical data to forecast future outcomes. Predicting next month's
sales from historical patterns is a predictive analytics task. Descriptive analytics would summarize past
sales, diagnostic would explain past performance, and prescriptive would recommend inventory levels.
Question 4:
What distinguishes supervised learning from unsupervised learning?
A) Supervised learning requires human intervention; unsupervised learning does not
B) Supervised learning uses labeled data (known outcomes); unsupervised learning uses unlabeled
data
C) Supervised learning is for prediction; unsupervised learning is for description only
D) Supervised learning is more accurate than unsupervised learning
Correct Answer: B
Rationale: Supervised learning algorithms require labeled training data where the target variable
(outcome) is known. Unsupervised learning works with unlabeled data to discover hidden patterns.
Both can be used for prediction and description, and accuracy depends on the problem.
Question 5:
Which of the following is a supervised learning task?
A) Discovering customer segments from purchase data
B) Identifying groups of similar documents
C) Predicting whether a loan application will default
D) Reducing dimensionality of image data
,Correct Answer: C
Rationale: Predicting loan default requires labeled historical data (defaulted or not) to train a
supervised model. Customer segmentation, document clustering, and dimensionality reduction are
unsupervised tasks that do not use pre-defined labels.
Question 6:
A healthcare provider wants to identify patients with similar disease profiles to develop targeted
treatment plans. This is an example of:
A) Classification
B) Regression
C) Clustering
D) Association rule mining
Correct Answer: C
Rationale: Grouping patients with similar disease profiles without predefined labels is clustering
(unsupervised learning). Classification would assign patients to known disease categories, regression
would predict continuous measures, and association rule mining would find relationships between
variables.
Question 7:
Association rule mining is best suited for which type of problem?
A) Predicting future sales
B) Finding items frequently purchased together
C) Segmenting customers
D) Forecasting stock prices
Correct Answer: B
Rationale: Association rule mining discovers relationships between variables, such as products
frequently purchased together (e.g., "customers who buy bread also buy butter"). It is commonly used
in market basket analysis. Prediction, segmentation, and forecasting require different methods.
Question 8:
The "curse of dimensionality" refers to:
A) The difficulty of visualizing high-dimensional data
B) The exponential increase in data volume needed to maintain statistical precision as dimensions
increase
, C) The computational cost of storing high-dimensional data
D) The inability of algorithms to handle categorical variables
Correct Answer: B
Rationale: The curse of dimensionality describes the exponential increase in data requirements as the
number of features grows. In high-dimensional spaces, data becomes sparse, distances become less
meaningful, and most points become equidistant, making many algorithms ineffective.
Question 9:
When the number of predictors (p) exceeds the number of observations (n), which method would be
most appropriate for prediction?
A) Ordinary least squares regression
B) Principal Component Analysis followed by regression
C) Simple linear regression with one predictor
D) No method can handle p > n
Correct Answer: B
Rationale: When p > n, OLS cannot provide unique coefficient estimates or suffers from overfitting.
PCA reduces dimensionality before regression, making the problem feasible. Feature selection
methods like LASSO are also appropriate.
Question 10:
Which of the following is a characteristic of big data described by the "Vs"?
A) Volume, Variety, Velocity
B) Volume, Variation, Verification
C) Variety, Validation, Value
D) Volume, Value, Visualization
Correct Answer: A
Rationale: The three Vs of big data are Volume (scale of data), Variety (different forms of data), and
Velocity (speed of data generation and processing). Additional Vs sometimes cited include Veracity
(data quality) and Value (business benefit).
Question 11:
What is the purpose of data splitting in predictive modeling?
Practice Examination
Comprehensive Mastery Assessment - Second Edition
Exam Title: ISYE 6501 Comprehensive Final Examination: Advanced Analytics Modeling Practice
Assessment Covering Supervised Learning, Unsupervised Learning, Time Series Analysis, Optimization
Methods, Simulation Techniques, Experimental Design, and Model Validation for Graduate-Level
Analytics Practitioners
SECTION 1: FUNDAMENTAL CONCEPTS AND ANALYTICS
FRAMEWORK (Questions 1-20)
Question 1:
In the analytics framework, which type of question addresses "What actions should be taken to achieve
desired outcomes?"
A) Descriptive analytics
B) Predictive analytics
C) Prescriptive analytics
D) Diagnostic analytics
Correct Answer: C
Rationale: Prescriptive analytics focuses on recommending actions and decisions to achieve specific
goals. It answers "what should we do?" by using optimization, simulation, and decision analysis to
guide decision-making. Descriptive analytics describes what happened, diagnostic analytics explains
why it happened, and predictive analytics forecasts what will happen.
Question 2:
Which of the following best represents a descriptive analytics question?
A) How many units should we produce next month?
B) Which customers are most likely to respond to our marketing campaign?
C) What were our total sales by region last quarter?
D) Why did sales decline in the Northeast region?
Correct Answer: C
,Rationale: Descriptive analytics summarizes historical data to understand what has happened. Asking
about total sales by region in the past quarter is a descriptive question. Predictive questions forecast
future outcomes, prescriptive questions recommend actions, and diagnostic questions explore root
causes.
Question 3:
A retail company wants to predict next month's sales based on historical data. This is an example of:
A) Descriptive analytics
B) Predictive analytics
C) Prescriptive analytics
D) Diagnostic analytics
Correct Answer: B
Rationale: Predictive analytics uses historical data to forecast future outcomes. Predicting next month's
sales from historical patterns is a predictive analytics task. Descriptive analytics would summarize past
sales, diagnostic would explain past performance, and prescriptive would recommend inventory levels.
Question 4:
What distinguishes supervised learning from unsupervised learning?
A) Supervised learning requires human intervention; unsupervised learning does not
B) Supervised learning uses labeled data (known outcomes); unsupervised learning uses unlabeled
data
C) Supervised learning is for prediction; unsupervised learning is for description only
D) Supervised learning is more accurate than unsupervised learning
Correct Answer: B
Rationale: Supervised learning algorithms require labeled training data where the target variable
(outcome) is known. Unsupervised learning works with unlabeled data to discover hidden patterns.
Both can be used for prediction and description, and accuracy depends on the problem.
Question 5:
Which of the following is a supervised learning task?
A) Discovering customer segments from purchase data
B) Identifying groups of similar documents
C) Predicting whether a loan application will default
D) Reducing dimensionality of image data
,Correct Answer: C
Rationale: Predicting loan default requires labeled historical data (defaulted or not) to train a
supervised model. Customer segmentation, document clustering, and dimensionality reduction are
unsupervised tasks that do not use pre-defined labels.
Question 6:
A healthcare provider wants to identify patients with similar disease profiles to develop targeted
treatment plans. This is an example of:
A) Classification
B) Regression
C) Clustering
D) Association rule mining
Correct Answer: C
Rationale: Grouping patients with similar disease profiles without predefined labels is clustering
(unsupervised learning). Classification would assign patients to known disease categories, regression
would predict continuous measures, and association rule mining would find relationships between
variables.
Question 7:
Association rule mining is best suited for which type of problem?
A) Predicting future sales
B) Finding items frequently purchased together
C) Segmenting customers
D) Forecasting stock prices
Correct Answer: B
Rationale: Association rule mining discovers relationships between variables, such as products
frequently purchased together (e.g., "customers who buy bread also buy butter"). It is commonly used
in market basket analysis. Prediction, segmentation, and forecasting require different methods.
Question 8:
The "curse of dimensionality" refers to:
A) The difficulty of visualizing high-dimensional data
B) The exponential increase in data volume needed to maintain statistical precision as dimensions
increase
, C) The computational cost of storing high-dimensional data
D) The inability of algorithms to handle categorical variables
Correct Answer: B
Rationale: The curse of dimensionality describes the exponential increase in data requirements as the
number of features grows. In high-dimensional spaces, data becomes sparse, distances become less
meaningful, and most points become equidistant, making many algorithms ineffective.
Question 9:
When the number of predictors (p) exceeds the number of observations (n), which method would be
most appropriate for prediction?
A) Ordinary least squares regression
B) Principal Component Analysis followed by regression
C) Simple linear regression with one predictor
D) No method can handle p > n
Correct Answer: B
Rationale: When p > n, OLS cannot provide unique coefficient estimates or suffers from overfitting.
PCA reduces dimensionality before regression, making the problem feasible. Feature selection
methods like LASSO are also appropriate.
Question 10:
Which of the following is a characteristic of big data described by the "Vs"?
A) Volume, Variety, Velocity
B) Volume, Variation, Verification
C) Variety, Validation, Value
D) Volume, Value, Visualization
Correct Answer: A
Rationale: The three Vs of big data are Volume (scale of data), Variety (different forms of data), and
Velocity (speed of data generation and processing). Additional Vs sometimes cited include Veracity
(data quality) and Value (business benefit).
Question 11:
What is the purpose of data splitting in predictive modeling?
