WGU D600 Exam 1: Linear Regression
Analysis | Complete Report, Python
Modeling, Results & Visualizations |
2026/2027 {JUST RELEASED} Updated
Practice Question Bank & Answer Rationale
THIS PACK CONTAINS
100 exam-style multiple-choice questions. Each question includes the correct answer and a complete
rationale for focused revision.
COVERAGE
This exam assesses mastery of simple linear regression, OLS estimation, assumption diagnostics, model
evaluation, Python implementation (statsmodels/scikit-learn/pandas), data preparation, and professional
reporting with visualizations. Students must interpret coefficients, diagnose violations, and communicate
findings in a complete analytical report.
STUDY GUIDE
Recommended duration: 3 hours. Passing target: 80%.
HIGHP - Page 1 of 32
,1. In the housing price regression, why is HouseColor as a categorical predictor
potentially problematic for OLS estimation?
[ ] A. It introduces perfect multicollinearity because colors are equally frequent.
[ ] B. It requires dummy coding; without reference category, it causes dummy variable trap.
[ ] C. It is a nominal variable and cannot be included in linear regression.
[ ] D. It has too many categories relative to sample size, causing overfitting.
CORRECT: B. It requires dummy coding; without reference category, it causes dummy variable trap.
Categorical variables like HouseColor must be dummy-coded; including all categories without dropping one
creates perfect multicollinearity (dummy variable trap). A is false because equal frequencies do not cause
perfect collinearity. C is false because nominal variables can be included via dummies. D is not necessarily
true; five categories are manageable.
2. The descriptive statistics show Fireplace: No=0.74, Yes=0.26. How should Fireplace be
encoded for regression?
[ ] A. As a numeric variable with values 0 and 1, treating 'Yes' as 1.
[ ] B. As two separate dummy variables for 'No' and 'Yes'.
[ ] C. As a factor with three levels to capture uncertainty.
[ ] D. Exclude it because it is imbalanced.
CORRECT: A. As a numeric variable with values 0 and 1, treating 'Yes' as 1.
Binary categorical variables are typically encoded as 0/1 dummy variables, with one level as reference. B
would create perfect collinearity. C is incorrect because there are only two levels. D is unnecessary;
imbalance does not preclude inclusion.
3. After fitting the model, residual plots show increasing variance with fitted values. Which
assumption is violated?
[ ] A. Linearity
[ ] B. Independence of errors
[ ] C. Homoscedasticity
[ ] D. Normality of errors
CORRECT: C. Homoscedasticity
A fan-shaped pattern in residuals versus fitted values indicates heteroscedasticity, violating the constant
variance (homoscedasticity) assumption. Linearity would show curvature, independence would show patterns
over time, and normality is assessed via Q-Q plots.
4. In Python statsmodels, which method provides a comprehensive regression summary
including coefficients, standard errors, and diagnostics?
[ ] A. model.fit().summary()
HIGHP - Page 2 of 32
, [ ] B. model.summary()
[ ] C. sm.OLS(y, X).fit().summary()
[ ] D. sm.add_constant(X).summary()
CORRECT: C. sm.OLS(y, X).fit().summary()
The correct workflow is to create an OLS model with sm.OLS(y, X), fit it, and call summary(). A and B are
incomplete or incorrect because model must be fitted. D is wrong because add_constant returns a
DataFrame, not a fitted model.
5. The model yields R-squared = 0.85. What does this indicate about the model's
explanatory power?
[ ] A. 85% of the variance in the dependent variable is explained by the predictors.
[ ] B. 85% of the predictions are correct.
[ ] C. The model explains 85% of the total sum of squares.
[ ] D. The model's RMSE is 0.85.
CORRECT: A. 85% of the variance in the dependent variable is explained by the predictors.
R-squared measures the proportion of variance in the dependent variable explained by the model. B
confuses with accuracy. C is a different formulation (R-squared = 1 - SSres/SStot). D is unrelated.
6. Which visualization is most appropriate to check the normality assumption of
residuals?
[ ] A. Histogram of residuals
[ ] B. Q-Q plot
[ ] C. Residuals vs. fitted values
[ ] D. Scale-location plot
CORRECT: B. Q-Q plot
A Q-Q plot directly compares residual quantiles to theoretical normal quantiles, making it the best for
assessing normality. Histograms are less precise, and the other plots assess different assumptions
(linearity/homoscedasticity).
7. When reporting regression results, which metric is preferred for comparing models with
different numbers of predictors?
[ ] A. R-squared
[ ] B. Adjusted R-squared
[ ] C. RMSE
[ ] D. p-values of coefficients
HIGHP - Page 3 of 32
, CORRECT: B. Adjusted R-squared
Adjusted R-squared penalizes for adding predictors, making it suitable for model comparison. R-squared
always increases with more predictors. RMSE is scale-dependent and not adjusted. p-values assess
individual significance, not overall model fit.
8. A residual vs. fitted plot shows a curved pattern. Which assumption is most likely
violated?
[ ] A. Normality
[ ] B. Homoscedasticity
[ ] C. Linearity
[ ] D. Independence
CORRECT: C. Linearity
A curved pattern indicates a non-linear relationship between predictors and outcome, violating the linearity
assumption. Homoscedasticity would show changing spread, normality would be assessed via Q-Q plot, and
independence via patterns over time or space.
9. In the context of the housing price model, which step ensures the data adheres to
regression assumptions before modeling?
[ ] A. Computing descriptive statistics and visualizations
[ ] B. Applying log transformation to the dependent variable
[ ] C. Removing outliers without justification
[ ] D. Using all available variables regardless of relevance
CORRECT: A. Computing descriptive statistics and visualizations
Descriptive statistics and visualizations help understand distributions, identify outliers, and check linearity,
ensuring data suitability. Log transformation may be a remedy but is not a preliminary step. Removing outliers
without justification is bad practice. Using all variables can introduce multicollinearity.
10. Which Git practice best supports traceability in a regression analysis project?
[ ] A. Committing after each rubric-aligned section with descriptive messages
[ ] B. Committing only at the end of the project
[ ] C. Using a single branch for all work
[ ] D. Storing data files in the repository without version control
CORRECT: A. Committing after each rubric-aligned section with descriptive messages
Frequent commits with descriptive messages after each section provide clear version history and traceability.
Committing only at the end loses granularity. A single branch may be fine but doesn't ensure traceability. Data
files should be versioned or handled separately.
HIGHP - Page 4 of 32
Analysis | Complete Report, Python
Modeling, Results & Visualizations |
2026/2027 {JUST RELEASED} Updated
Practice Question Bank & Answer Rationale
THIS PACK CONTAINS
100 exam-style multiple-choice questions. Each question includes the correct answer and a complete
rationale for focused revision.
COVERAGE
This exam assesses mastery of simple linear regression, OLS estimation, assumption diagnostics, model
evaluation, Python implementation (statsmodels/scikit-learn/pandas), data preparation, and professional
reporting with visualizations. Students must interpret coefficients, diagnose violations, and communicate
findings in a complete analytical report.
STUDY GUIDE
Recommended duration: 3 hours. Passing target: 80%.
HIGHP - Page 1 of 32
,1. In the housing price regression, why is HouseColor as a categorical predictor
potentially problematic for OLS estimation?
[ ] A. It introduces perfect multicollinearity because colors are equally frequent.
[ ] B. It requires dummy coding; without reference category, it causes dummy variable trap.
[ ] C. It is a nominal variable and cannot be included in linear regression.
[ ] D. It has too many categories relative to sample size, causing overfitting.
CORRECT: B. It requires dummy coding; without reference category, it causes dummy variable trap.
Categorical variables like HouseColor must be dummy-coded; including all categories without dropping one
creates perfect multicollinearity (dummy variable trap). A is false because equal frequencies do not cause
perfect collinearity. C is false because nominal variables can be included via dummies. D is not necessarily
true; five categories are manageable.
2. The descriptive statistics show Fireplace: No=0.74, Yes=0.26. How should Fireplace be
encoded for regression?
[ ] A. As a numeric variable with values 0 and 1, treating 'Yes' as 1.
[ ] B. As two separate dummy variables for 'No' and 'Yes'.
[ ] C. As a factor with three levels to capture uncertainty.
[ ] D. Exclude it because it is imbalanced.
CORRECT: A. As a numeric variable with values 0 and 1, treating 'Yes' as 1.
Binary categorical variables are typically encoded as 0/1 dummy variables, with one level as reference. B
would create perfect collinearity. C is incorrect because there are only two levels. D is unnecessary;
imbalance does not preclude inclusion.
3. After fitting the model, residual plots show increasing variance with fitted values. Which
assumption is violated?
[ ] A. Linearity
[ ] B. Independence of errors
[ ] C. Homoscedasticity
[ ] D. Normality of errors
CORRECT: C. Homoscedasticity
A fan-shaped pattern in residuals versus fitted values indicates heteroscedasticity, violating the constant
variance (homoscedasticity) assumption. Linearity would show curvature, independence would show patterns
over time, and normality is assessed via Q-Q plots.
4. In Python statsmodels, which method provides a comprehensive regression summary
including coefficients, standard errors, and diagnostics?
[ ] A. model.fit().summary()
HIGHP - Page 2 of 32
, [ ] B. model.summary()
[ ] C. sm.OLS(y, X).fit().summary()
[ ] D. sm.add_constant(X).summary()
CORRECT: C. sm.OLS(y, X).fit().summary()
The correct workflow is to create an OLS model with sm.OLS(y, X), fit it, and call summary(). A and B are
incomplete or incorrect because model must be fitted. D is wrong because add_constant returns a
DataFrame, not a fitted model.
5. The model yields R-squared = 0.85. What does this indicate about the model's
explanatory power?
[ ] A. 85% of the variance in the dependent variable is explained by the predictors.
[ ] B. 85% of the predictions are correct.
[ ] C. The model explains 85% of the total sum of squares.
[ ] D. The model's RMSE is 0.85.
CORRECT: A. 85% of the variance in the dependent variable is explained by the predictors.
R-squared measures the proportion of variance in the dependent variable explained by the model. B
confuses with accuracy. C is a different formulation (R-squared = 1 - SSres/SStot). D is unrelated.
6. Which visualization is most appropriate to check the normality assumption of
residuals?
[ ] A. Histogram of residuals
[ ] B. Q-Q plot
[ ] C. Residuals vs. fitted values
[ ] D. Scale-location plot
CORRECT: B. Q-Q plot
A Q-Q plot directly compares residual quantiles to theoretical normal quantiles, making it the best for
assessing normality. Histograms are less precise, and the other plots assess different assumptions
(linearity/homoscedasticity).
7. When reporting regression results, which metric is preferred for comparing models with
different numbers of predictors?
[ ] A. R-squared
[ ] B. Adjusted R-squared
[ ] C. RMSE
[ ] D. p-values of coefficients
HIGHP - Page 3 of 32
, CORRECT: B. Adjusted R-squared
Adjusted R-squared penalizes for adding predictors, making it suitable for model comparison. R-squared
always increases with more predictors. RMSE is scale-dependent and not adjusted. p-values assess
individual significance, not overall model fit.
8. A residual vs. fitted plot shows a curved pattern. Which assumption is most likely
violated?
[ ] A. Normality
[ ] B. Homoscedasticity
[ ] C. Linearity
[ ] D. Independence
CORRECT: C. Linearity
A curved pattern indicates a non-linear relationship between predictors and outcome, violating the linearity
assumption. Homoscedasticity would show changing spread, normality would be assessed via Q-Q plot, and
independence via patterns over time or space.
9. In the context of the housing price model, which step ensures the data adheres to
regression assumptions before modeling?
[ ] A. Computing descriptive statistics and visualizations
[ ] B. Applying log transformation to the dependent variable
[ ] C. Removing outliers without justification
[ ] D. Using all available variables regardless of relevance
CORRECT: A. Computing descriptive statistics and visualizations
Descriptive statistics and visualizations help understand distributions, identify outliers, and check linearity,
ensuring data suitability. Log transformation may be a remedy but is not a preliminary step. Removing outliers
without justification is bad practice. Using all variables can introduce multicollinearity.
10. Which Git practice best supports traceability in a regression analysis project?
[ ] A. Committing after each rubric-aligned section with descriptive messages
[ ] B. Committing only at the end of the project
[ ] C. Using a single branch for all work
[ ] D. Storing data files in the repository without version control
CORRECT: A. Committing after each rubric-aligned section with descriptive messages
Frequent commits with descriptive messages after each section provide clear version history and traceability.
Committing only at the end loses granularity. A single branch may be fine but doesn't ensure traceability. Data
files should be versioned or handled separately.
HIGHP - Page 4 of 32
