GEORGIA TECH ISYE 6402 MIDTERM 1 EXAM :ANALYSIS
OF DOMESTIC PASSENGER DATA | QUESTIONS AND
ANSWERS | 2026 UPDATED | 100% CORRECT.
109 Questions with Answers and Detailed Rationales
100 PERCENT GUARANTEED PASS
INSTANT DOWNLOAD ANSWERS INCLUDED
IMPORTANCE OF THIS DOCUMENT
This comprehensive examination preparation guide has been meticulously developed to help you succeed in the
GEORGIA TECH ISYE 6402 MIDTERM 1 EXAM :ANALYSIS OF DOMESTIC PASSENGER DATA | QUESTIONS
AND ANSWERS | 2026 UPDATED | 100% CORRECT.. It contains 109 carefully selected questions that reflect
the most current exam content and testing strategies. Each question is accompanied by a correct answer and a
detailed rationale that explains the underlying pathophysiology, pharmacology, or clinical reasoning.
Self-Assessment – Test your knowledge and Exam Preparation – Familiarize yourself with the
identify areas requiring further question format and content
study areas
Concept Reinforcement – Deepen your Confidence Building – Develop test-taking
understanding through strategies and reduce
evidence-based exam anxiety
rationales
Time Management – Practice answering
questions under simulated
exam conditions
Review Summary 109 Questions
Foundations - Application - Georgia TECH ISYE 6402 1 Analysis OF Domestic Passenger DATA AND 2026
Updated 100 Correct DATA Science / Statistics / Transportation Analytics Graduate
All answers with rationales
,Table of Contents
Content Area Questions Key Topics
DATA Exploration AND 1-19 Passenger, Regression, Model, Predicting, Appropriate
Visualization
Linear Regression Models 20-38 Model, Passenger, Regression, Domestic, Distance
Model Diagnostics AND 39-57 Model, Regression, Coefficient, Passenger, Airline
Residual Analysis
TIME Series Analysis AND 58-76 Model, Passenger, Domestic, Regression, Significant
Forecasting
Statistical Inference AND 77-95 Model, Domestic, Passenger, Appropriate, Regression
Hypothesis Testing
Anova AND Experimental 96-109 Passenger, Model, Regression, Domestic, Variance
Design
TOTAL 109 All questions include answers and detailed rationales
,Section A - DATA Exploration AND Visualization
Q1.
A data scientist fits a seasonal ARIMA model to monthly domestic passenger counts. The
ACF of the residuals shows a significant spike at lag 12, while the PACF decays. What
should be the next step?
A. Add a seasonal MA(1) term to the model B. Add a seasonal AR(1) term to the model
C. Increase the differencing order d D. Reduce the non-seasonal MA order
Correct: A - Add a seasonal MA(1) term to the model
Rationale:A significant spike at lag 12 in the ACF with PACF decay suggests an unmodeled
seasonal MA component. Adding a seasonal MA(1) term is appropriate. A seasonal AR term
would show a decaying ACF and spike in PACF. Differencing order is about stationarity, not
residual correlation. Reducing non-seasonal MA would not address the seasonal pattern.
Q2.
When comparing two nested regression models for predicting passenger volume, which
approach correctly accounts for the number of parameters?
A. Use the F-test based on the difference in B. Compare AIC values, lower is better
R-squared
C. Use the likelihood ratio test with D. Select the model with the higher adjusted
chi-square degrees of freedom equal to the R-squared
difference in parameters
Correct: C - Use the likelihood ratio test with chi-square degrees of freedom equal to the
difference in parameters
Rationale:The likelihood ratio test formally compares nested models, with the test statistic
following a chi-square distribution with degrees of freedom equal to the difference in the
number of parameters. The F-test is equivalent but uses an F-distribution. AIC and adjusted
R-squared are model selection criteria but do not provide a formal hypothesis test. The F-test
on R-squared difference is not standard.
Q3.
In a logistic regression model predicting flight cancellation, you encounter quasi-complete
separation. Which approach is most appropriate?
A. Remove the offending predictor B. Apply Firth's penalized likelihood method
C. Increase the sample size by D. Use ordinary least squares on the binary
bootstrapping outcome
Page 3
, Section A - DATA Exploration AND Visualization
Correct: B - Apply Firth's penalized likelihood method
Rationale:Firth's penalized likelihood is designed to handle separation by reducing bias in
maximum likelihood estimates. Removing the predictor loses information. Bootstrapping does
not solve the underlying separation issue. OLS is inappropriate for binary outcomes due to
heteroscedasticity and predictions outside [0,1].
Q4.
You have a time series of daily passenger counts with missing values. To perform ARIMA
modeling, which imputation method is least likely to introduce bias in the autocorrelation
structure?
A. Mean imputation B. Last observation carried forward
C. Linear interpolation D. Kalman filter based imputation using
state-space representation
Correct: D - Kalman filter based imputation using state-space representation
Rationale:Kalman filter imputation uses the underlying state-space model, preserving the
time series dynamics and uncertainty. Mean imputation and LOCF distort autocorrelation.
Linear interpolation assumes a smooth trend and can dampen short-term fluctuations, biasing
estimates of autocorrelation.
Q5.
When visualizing high-dimensional passenger data, which method is best for preserving
global structure while allowing nonlinear relationships?
A. Principal Component Analysis (PCA) B. t-Distributed Stochastic Neighbor
Embedding (t-SNE)
C. Multidimensional Scaling (MDS) D. Uniform Manifold Approximation and
Projection (UMAP)
Correct: D - Uniform Manifold Approximation and Projection (UMAP)
Rationale:UMAP balances preserving both local and global structure, making it superior for
high-dimensional data with nonlinear relationships. PCA is linear and only captures variance.
t-SNE focuses on local structure and often distorts global distances. Classical MDS is linear
and assumes Euclidean distances.
Q6.
In a multiple linear regression predicting passenger demand, you detect
heteroscedasticity in the residuals. What is the most appropriate remedy?
A. Apply a Box-Cox transformation to the B. Use robust standard errors (Huber-White)
response variable
Page 4
OF DOMESTIC PASSENGER DATA | QUESTIONS AND
ANSWERS | 2026 UPDATED | 100% CORRECT.
109 Questions with Answers and Detailed Rationales
100 PERCENT GUARANTEED PASS
INSTANT DOWNLOAD ANSWERS INCLUDED
IMPORTANCE OF THIS DOCUMENT
This comprehensive examination preparation guide has been meticulously developed to help you succeed in the
GEORGIA TECH ISYE 6402 MIDTERM 1 EXAM :ANALYSIS OF DOMESTIC PASSENGER DATA | QUESTIONS
AND ANSWERS | 2026 UPDATED | 100% CORRECT.. It contains 109 carefully selected questions that reflect
the most current exam content and testing strategies. Each question is accompanied by a correct answer and a
detailed rationale that explains the underlying pathophysiology, pharmacology, or clinical reasoning.
Self-Assessment – Test your knowledge and Exam Preparation – Familiarize yourself with the
identify areas requiring further question format and content
study areas
Concept Reinforcement – Deepen your Confidence Building – Develop test-taking
understanding through strategies and reduce
evidence-based exam anxiety
rationales
Time Management – Practice answering
questions under simulated
exam conditions
Review Summary 109 Questions
Foundations - Application - Georgia TECH ISYE 6402 1 Analysis OF Domestic Passenger DATA AND 2026
Updated 100 Correct DATA Science / Statistics / Transportation Analytics Graduate
All answers with rationales
,Table of Contents
Content Area Questions Key Topics
DATA Exploration AND 1-19 Passenger, Regression, Model, Predicting, Appropriate
Visualization
Linear Regression Models 20-38 Model, Passenger, Regression, Domestic, Distance
Model Diagnostics AND 39-57 Model, Regression, Coefficient, Passenger, Airline
Residual Analysis
TIME Series Analysis AND 58-76 Model, Passenger, Domestic, Regression, Significant
Forecasting
Statistical Inference AND 77-95 Model, Domestic, Passenger, Appropriate, Regression
Hypothesis Testing
Anova AND Experimental 96-109 Passenger, Model, Regression, Domestic, Variance
Design
TOTAL 109 All questions include answers and detailed rationales
,Section A - DATA Exploration AND Visualization
Q1.
A data scientist fits a seasonal ARIMA model to monthly domestic passenger counts. The
ACF of the residuals shows a significant spike at lag 12, while the PACF decays. What
should be the next step?
A. Add a seasonal MA(1) term to the model B. Add a seasonal AR(1) term to the model
C. Increase the differencing order d D. Reduce the non-seasonal MA order
Correct: A - Add a seasonal MA(1) term to the model
Rationale:A significant spike at lag 12 in the ACF with PACF decay suggests an unmodeled
seasonal MA component. Adding a seasonal MA(1) term is appropriate. A seasonal AR term
would show a decaying ACF and spike in PACF. Differencing order is about stationarity, not
residual correlation. Reducing non-seasonal MA would not address the seasonal pattern.
Q2.
When comparing two nested regression models for predicting passenger volume, which
approach correctly accounts for the number of parameters?
A. Use the F-test based on the difference in B. Compare AIC values, lower is better
R-squared
C. Use the likelihood ratio test with D. Select the model with the higher adjusted
chi-square degrees of freedom equal to the R-squared
difference in parameters
Correct: C - Use the likelihood ratio test with chi-square degrees of freedom equal to the
difference in parameters
Rationale:The likelihood ratio test formally compares nested models, with the test statistic
following a chi-square distribution with degrees of freedom equal to the difference in the
number of parameters. The F-test is equivalent but uses an F-distribution. AIC and adjusted
R-squared are model selection criteria but do not provide a formal hypothesis test. The F-test
on R-squared difference is not standard.
Q3.
In a logistic regression model predicting flight cancellation, you encounter quasi-complete
separation. Which approach is most appropriate?
A. Remove the offending predictor B. Apply Firth's penalized likelihood method
C. Increase the sample size by D. Use ordinary least squares on the binary
bootstrapping outcome
Page 3
, Section A - DATA Exploration AND Visualization
Correct: B - Apply Firth's penalized likelihood method
Rationale:Firth's penalized likelihood is designed to handle separation by reducing bias in
maximum likelihood estimates. Removing the predictor loses information. Bootstrapping does
not solve the underlying separation issue. OLS is inappropriate for binary outcomes due to
heteroscedasticity and predictions outside [0,1].
Q4.
You have a time series of daily passenger counts with missing values. To perform ARIMA
modeling, which imputation method is least likely to introduce bias in the autocorrelation
structure?
A. Mean imputation B. Last observation carried forward
C. Linear interpolation D. Kalman filter based imputation using
state-space representation
Correct: D - Kalman filter based imputation using state-space representation
Rationale:Kalman filter imputation uses the underlying state-space model, preserving the
time series dynamics and uncertainty. Mean imputation and LOCF distort autocorrelation.
Linear interpolation assumes a smooth trend and can dampen short-term fluctuations, biasing
estimates of autocorrelation.
Q5.
When visualizing high-dimensional passenger data, which method is best for preserving
global structure while allowing nonlinear relationships?
A. Principal Component Analysis (PCA) B. t-Distributed Stochastic Neighbor
Embedding (t-SNE)
C. Multidimensional Scaling (MDS) D. Uniform Manifold Approximation and
Projection (UMAP)
Correct: D - Uniform Manifold Approximation and Projection (UMAP)
Rationale:UMAP balances preserving both local and global structure, making it superior for
high-dimensional data with nonlinear relationships. PCA is linear and only captures variance.
t-SNE focuses on local structure and often distorts global distances. Classical MDS is linear
and assumes Euclidean distances.
Q6.
In a multiple linear regression predicting passenger demand, you detect
heteroscedasticity in the residuals. What is the most appropriate remedy?
A. Apply a Box-Cox transformation to the B. Use robust standard errors (Huber-White)
response variable
Page 4
