ISYE 6414 MIDTERM EXAM 1 & 2
COMPLETE BUNDLE | MULTIPLE
VERSIONS, QUESTIONS & ANSWERS, R
CODE AND DATA ANALYSIS SOLUTIONS |
2026/2027 UPDATED – GEORGIA TECH
| QUESTIONS 1-100 | DETAILED
RATIONALES
Study-use note: This is an original, advanced practice question bank designed around
common ISYE 6414 Regression Analysis concepts, including regression modeling,
inference, diagnostics, transformations, model selection, interactions, ANOVA, and R-
based analysis. It is not an official or confidential Georgia Tech exam, and no grade
or pass result is guaranteed.
INTRODUCTION
ISYE 6414 Regression Analysis focuses on the theory and practical application of
regression methods for explaining relationships between variables, making
predictions, assessing uncertainty, and diagnosing model adequacy. The course
requires students to move beyond simply fitting a regression equation: they must
understand assumptions, interpret coefficients correctly, construct and evaluate
models, recognize violations, and use statistical evidence to support defensible
conclusions. This practice bank is designed for students preparing for Midterm 1 and
Midterm 2 and emphasizes difficult application-level scenarios rather than simple
terminology. Questions incorporate multiple regression, least squares estimation,
matrix concepts, hypothesis testing, confidence and prediction intervals, categorical
predictors, interactions, polynomial terms, transformations, residual diagnostics,
multicollinearity, leverage and influence, model selection, ANOVA, and practical R-
based interpretation. The questions are intentionally structured to require careful
reasoning about what the fitted model actually implies. The accompanying rationales
explain both the correct reasoning and why plausible alternatives fail, helping students
identify common exam traps and strengthen their statistical judgment.
CORE DOMAINS TESTED
1.
Simple Linear Regression — least squares estimation, fitted values, residuals,
slopes, intercepts, and interpretation.
1
, 2.
3.
Multiple Linear Regression — partial effects, coefficient interpretation,
matrix formulation, and model fitting.
4.
5.
Statistical Inference — t-tests, F-tests, confidence intervals, p-values, and
hypothesis testing.
6.
7.
Model Assumptions — linearity, independence, constant variance, normality,
and consequences of violations.
8.
9.
ANOVA & Nested Models — decomposition of variation, partial F-tests, and
model comparison.
10.
11.
Categorical Predictors — indicator variables, reference groups, and
interpretation.
12.
13.
Interactions — effect modification and interpretation of interaction
coefficients.
14.
15.
Polynomial Regression & Transformations — nonlinear relationships, log
transformations, and interpretation.
16.
17.
Regression Diagnostics — residuals, leverage, influence, Cook's distance,
and outliers.
18.
2
, 19.
Multicollinearity — correlation among predictors, VIF, coefficient instability,
and interpretation.
20.
21.
Model Selection — AIC, adjusted R2R^2, variable selection, and predictive
considerations.
22.
23.
R & Data Analysis — interpreting regression output, diagnostic plots,
formulas, prediction intervals, and practical analysis.
24.
QUESTIONS 1-100
Q1: A researcher fits the model Yi=β0+β
1Xi+ϵiY_i=\beta_0+\beta_1X_i+\epsilon_i and obtains β
^1=3.7\hat{\beta}_1=3.7. Which interpretation is most
appropriate?
A) Increasing YY by one unit causes XX to increase by 3.7 units.
B) For a one-unit increase in XX, the fitted mean of YY increases by 3.7 units.
C) Every individual observation of YY increases by exactly 3.7 units.
D) The correlation between XX and YY is 3.7.
Rationale: Option B correctly interprets the regression slope as the change in the
conditional mean of YY associated with a one-unit increase in XX. Option A reverses
the roles of the variables. Option C incorrectly suggests that every individual
observation changes deterministically, ignoring residual variation. Option D is
impossible because correlation must lie between -1 and 1.
Q2: In ordinary least squares regression, what quantity is
minimized when estimating the regression coefficients?
A) Sum of absolute residuals
B) Sum of squared residuals
C) Sum of fitted values
D) Sum of predictor values
3
COMPLETE BUNDLE | MULTIPLE
VERSIONS, QUESTIONS & ANSWERS, R
CODE AND DATA ANALYSIS SOLUTIONS |
2026/2027 UPDATED – GEORGIA TECH
| QUESTIONS 1-100 | DETAILED
RATIONALES
Study-use note: This is an original, advanced practice question bank designed around
common ISYE 6414 Regression Analysis concepts, including regression modeling,
inference, diagnostics, transformations, model selection, interactions, ANOVA, and R-
based analysis. It is not an official or confidential Georgia Tech exam, and no grade
or pass result is guaranteed.
INTRODUCTION
ISYE 6414 Regression Analysis focuses on the theory and practical application of
regression methods for explaining relationships between variables, making
predictions, assessing uncertainty, and diagnosing model adequacy. The course
requires students to move beyond simply fitting a regression equation: they must
understand assumptions, interpret coefficients correctly, construct and evaluate
models, recognize violations, and use statistical evidence to support defensible
conclusions. This practice bank is designed for students preparing for Midterm 1 and
Midterm 2 and emphasizes difficult application-level scenarios rather than simple
terminology. Questions incorporate multiple regression, least squares estimation,
matrix concepts, hypothesis testing, confidence and prediction intervals, categorical
predictors, interactions, polynomial terms, transformations, residual diagnostics,
multicollinearity, leverage and influence, model selection, ANOVA, and practical R-
based interpretation. The questions are intentionally structured to require careful
reasoning about what the fitted model actually implies. The accompanying rationales
explain both the correct reasoning and why plausible alternatives fail, helping students
identify common exam traps and strengthen their statistical judgment.
CORE DOMAINS TESTED
1.
Simple Linear Regression — least squares estimation, fitted values, residuals,
slopes, intercepts, and interpretation.
1
, 2.
3.
Multiple Linear Regression — partial effects, coefficient interpretation,
matrix formulation, and model fitting.
4.
5.
Statistical Inference — t-tests, F-tests, confidence intervals, p-values, and
hypothesis testing.
6.
7.
Model Assumptions — linearity, independence, constant variance, normality,
and consequences of violations.
8.
9.
ANOVA & Nested Models — decomposition of variation, partial F-tests, and
model comparison.
10.
11.
Categorical Predictors — indicator variables, reference groups, and
interpretation.
12.
13.
Interactions — effect modification and interpretation of interaction
coefficients.
14.
15.
Polynomial Regression & Transformations — nonlinear relationships, log
transformations, and interpretation.
16.
17.
Regression Diagnostics — residuals, leverage, influence, Cook's distance,
and outliers.
18.
2
, 19.
Multicollinearity — correlation among predictors, VIF, coefficient instability,
and interpretation.
20.
21.
Model Selection — AIC, adjusted R2R^2, variable selection, and predictive
considerations.
22.
23.
R & Data Analysis — interpreting regression output, diagnostic plots,
formulas, prediction intervals, and practical analysis.
24.
QUESTIONS 1-100
Q1: A researcher fits the model Yi=β0+β
1Xi+ϵiY_i=\beta_0+\beta_1X_i+\epsilon_i and obtains β
^1=3.7\hat{\beta}_1=3.7. Which interpretation is most
appropriate?
A) Increasing YY by one unit causes XX to increase by 3.7 units.
B) For a one-unit increase in XX, the fitted mean of YY increases by 3.7 units.
C) Every individual observation of YY increases by exactly 3.7 units.
D) The correlation between XX and YY is 3.7.
Rationale: Option B correctly interprets the regression slope as the change in the
conditional mean of YY associated with a one-unit increase in XX. Option A reverses
the roles of the variables. Option C incorrectly suggests that every individual
observation changes deterministically, ignoring residual variation. Option D is
impossible because correlation must lie between -1 and 1.
Q2: In ordinary least squares regression, what quantity is
minimized when estimating the regression coefficients?
A) Sum of absolute residuals
B) Sum of squared residuals
C) Sum of fitted values
D) Sum of predictor values
3
