ISYE 6501 Final Exam Prep 2026 | 200 Practice Questions with
Answers & Explanations | Data Analytics Study Guide
Question Types: True/False (TF) & Multiple Choice (MC)
Total Questions: 200
Topics: Classification, Clustering, Regression, Time Series, Optimization, Simulation, Data Preparation,
Model Evaluation
Spring 2026 — 100% Correct Answers with Explanations
Section 1: Data Preparation & Exploratory Analysis (Questions 1–20)
1. In a boxplot, the “whiskers” typically extend to the minimum and maximum values in the data set.
Answer: False
Explanation: By default, whiskers extend to the most extreme data points that are not outliers (usually
within 1.5 × IQR from the quartiles). Outliers are plotted individually.
2. A histogram is used to visualize the distribution of a continuous variable.
Answer: True
Explanation: Histograms group continuous data into bins and display frequencies, revealing shape,
central tendency, and spread.
3. The interquartile range (IQR) is defined as Q3 – Q1 and is a measure of statistical dispersion.
Answer: True
Explanation: IQR represents the range of the middle 50% of the data. It is robust to outliers.
4. Removing outliers based on the “3‑sigma rule” is always safe and improves model performance.
Answer: False
,Explanation: Outliers may contain valuable information. Removing them arbitrarily can bias results. The
decision should be based on domain knowledge and model sensitivity.
5. Standardizing features (subtracting mean, dividing by standard deviation) changes the shape of the
distribution.
Answer: False
Explanation: Standardization shifts and scales the data but does not alter the shape (skewness, kurtosis).
The relative distances remain proportional.
6. Min‑max scaling transforms features to a fixed range, typically [0, 1].
Answer: True
Explanation: Min‑max scaling is defined as (x – min)/(max – min). It preserves the shape but can be
sensitive to outliers.
7. Missing data can be handled by mean imputation, but this may reduce the variance of the imputed
variable.
Answer: True
Explanation: Mean imputation artificially reduces variability and may distort correlations. More
advanced methods (e.g., multiple imputation) are often preferred.
8. A scatterplot matrix is useful for visualizing pairwise relationships among several continuous variables.
Answer: True
Explanation: Scatterplot matrices display all bivariate scatterplots in a grid, helping to detect correlations
and patterns.
9. A Q‑Q plot compares the quantiles of a sample to the quantiles of a theoretical distribution.
Answer: True
Explanation: If the points lie approximately on a straight line, the sample follows the chosen distribution.
10. Logarithmic transformation is often applied to right‑skewed data to make the distribution more
symmetric.
Answer: True
,Explanation: Log transformation compresses the right tail and can stabilize variance, making the data
closer to normal.
11. The correlation coefficient measures the strength of any relationship, not just linear.
Answer: False
Explanation: Pearson correlation measures only linear relationships. Variables with a strong quadratic or
other non‑linear relationship may have near‑zero correlation.
12. A high correlation between two variables implies causation.
Answer: False
Explanation: Correlation does not imply causation. There may be a lurking variable or reverse causality.
13. Principal component analysis (PCA) is a dimensionality reduction technique that creates new
uncorrelated features.
Answer: True
Explanation: PCA finds orthogonal linear combinations that maximize variance, reducing dimensionality
while preserving most information.
14. PCA can be used before clustering to reduce noise and speed up computation.
Answer: True
Explanation: PCA can remove low‑variance components (noise) and reduce the number of dimensions,
often improving clustering performance.
15. The first principal component explains the largest possible variance in the data.
Answer: True
Explanation: PCA sequentially finds directions of maximum variance, with the first PC accounting for the
greatest variance.
16. Before applying PCA, it is recommended to standardize the variables to have zero mean and unit
variance.
Answer: True
, Explanation: PCA is scale‑sensitive. Standardization ensures that variables with large scales do not
dominate the principal components.
17. The scree plot shows the eigenvalues or proportion of variance explained by each principal
component.
Answer: True
Explanation: A scree plot helps decide how many PCs to retain by looking for an “elbow” where
eigenvalues drop sharply.
18. Outliers can severely affect PCA results.
Answer: True
Explanation: Outliers can distort the covariance matrix, causing principal components to be pulled
toward the outliers. Robust PCA methods exist.
19. The covariance matrix is used in PCA when variables are on the same scale; otherwise the correlation
matrix is preferred.
Answer: True
Explanation: Using the covariance matrix is equivalent to PCA on unstandardized data. Using the
correlation matrix standardizes variables automatically.
20. PCA creates features that are linear combinations of the original variables and are correlated with
each other.
Answer: False
Explanation: Principal components are uncorrelated (orthogonal) by construction.
Section 2: Linear Regression & Regularization (Questions 21–40)
21. In linear regression, the least squares estimator minimizes the sum of squared residuals.
Answer: True
Answers & Explanations | Data Analytics Study Guide
Question Types: True/False (TF) & Multiple Choice (MC)
Total Questions: 200
Topics: Classification, Clustering, Regression, Time Series, Optimization, Simulation, Data Preparation,
Model Evaluation
Spring 2026 — 100% Correct Answers with Explanations
Section 1: Data Preparation & Exploratory Analysis (Questions 1–20)
1. In a boxplot, the “whiskers” typically extend to the minimum and maximum values in the data set.
Answer: False
Explanation: By default, whiskers extend to the most extreme data points that are not outliers (usually
within 1.5 × IQR from the quartiles). Outliers are plotted individually.
2. A histogram is used to visualize the distribution of a continuous variable.
Answer: True
Explanation: Histograms group continuous data into bins and display frequencies, revealing shape,
central tendency, and spread.
3. The interquartile range (IQR) is defined as Q3 – Q1 and is a measure of statistical dispersion.
Answer: True
Explanation: IQR represents the range of the middle 50% of the data. It is robust to outliers.
4. Removing outliers based on the “3‑sigma rule” is always safe and improves model performance.
Answer: False
,Explanation: Outliers may contain valuable information. Removing them arbitrarily can bias results. The
decision should be based on domain knowledge and model sensitivity.
5. Standardizing features (subtracting mean, dividing by standard deviation) changes the shape of the
distribution.
Answer: False
Explanation: Standardization shifts and scales the data but does not alter the shape (skewness, kurtosis).
The relative distances remain proportional.
6. Min‑max scaling transforms features to a fixed range, typically [0, 1].
Answer: True
Explanation: Min‑max scaling is defined as (x – min)/(max – min). It preserves the shape but can be
sensitive to outliers.
7. Missing data can be handled by mean imputation, but this may reduce the variance of the imputed
variable.
Answer: True
Explanation: Mean imputation artificially reduces variability and may distort correlations. More
advanced methods (e.g., multiple imputation) are often preferred.
8. A scatterplot matrix is useful for visualizing pairwise relationships among several continuous variables.
Answer: True
Explanation: Scatterplot matrices display all bivariate scatterplots in a grid, helping to detect correlations
and patterns.
9. A Q‑Q plot compares the quantiles of a sample to the quantiles of a theoretical distribution.
Answer: True
Explanation: If the points lie approximately on a straight line, the sample follows the chosen distribution.
10. Logarithmic transformation is often applied to right‑skewed data to make the distribution more
symmetric.
Answer: True
,Explanation: Log transformation compresses the right tail and can stabilize variance, making the data
closer to normal.
11. The correlation coefficient measures the strength of any relationship, not just linear.
Answer: False
Explanation: Pearson correlation measures only linear relationships. Variables with a strong quadratic or
other non‑linear relationship may have near‑zero correlation.
12. A high correlation between two variables implies causation.
Answer: False
Explanation: Correlation does not imply causation. There may be a lurking variable or reverse causality.
13. Principal component analysis (PCA) is a dimensionality reduction technique that creates new
uncorrelated features.
Answer: True
Explanation: PCA finds orthogonal linear combinations that maximize variance, reducing dimensionality
while preserving most information.
14. PCA can be used before clustering to reduce noise and speed up computation.
Answer: True
Explanation: PCA can remove low‑variance components (noise) and reduce the number of dimensions,
often improving clustering performance.
15. The first principal component explains the largest possible variance in the data.
Answer: True
Explanation: PCA sequentially finds directions of maximum variance, with the first PC accounting for the
greatest variance.
16. Before applying PCA, it is recommended to standardize the variables to have zero mean and unit
variance.
Answer: True
, Explanation: PCA is scale‑sensitive. Standardization ensures that variables with large scales do not
dominate the principal components.
17. The scree plot shows the eigenvalues or proportion of variance explained by each principal
component.
Answer: True
Explanation: A scree plot helps decide how many PCs to retain by looking for an “elbow” where
eigenvalues drop sharply.
18. Outliers can severely affect PCA results.
Answer: True
Explanation: Outliers can distort the covariance matrix, causing principal components to be pulled
toward the outliers. Robust PCA methods exist.
19. The covariance matrix is used in PCA when variables are on the same scale; otherwise the correlation
matrix is preferred.
Answer: True
Explanation: Using the covariance matrix is equivalent to PCA on unstandardized data. Using the
correlation matrix standardizes variables automatically.
20. PCA creates features that are linear combinations of the original variables and are correlated with
each other.
Answer: False
Explanation: Principal components are uncorrelated (orthogonal) by construction.
Section 2: Linear Regression & Regularization (Questions 21–40)
21. In linear regression, the least squares estimator minimizes the sum of squared residuals.
Answer: True
