Senior Data Scientist (SDS) Practice Exam 2026 Edition
Q1. Which of the following best describes supervised learning?
A. Learning patterns without labeled outputs
B. Learning patterns with labeled outputs
C. Only clustering data
D. Only dimensionality reduction
Answer: B
Explanation: Supervised learning trains models on labeled data to predict outcomes.
Q2. Unsupervised learning is primarily used for:
A. Regression
B. Classification
C. Clustering and pattern discovery
D. Time series forecasting
Answer: C
Explanation: Unsupervised learning finds hidden structures without labeled outputs.
Q3. Which of the following is a continuous variable?
A. Number of cars
B. Temperature in Celsius
C. Gender
D. Country
Answer: B
Explanation: Continuous variables can take any real value within a range.
Q4. In data science, bias refers to:
A. Systematic error in predictions
B. Random noise
C. Sample size
D. Data visualization choice
Answer: A
Explanation: Bias causes a model to consistently underperform in certain areas.
Q5. Variance in a model measures:
A. How far predictions are from the mean
B. Consistency of predictions across datasets
C. Both bias and error
D. Only model accuracy
Answer: B
Explanation: High variance indicates overfitting to training data.
,Senior Data Scientist (SDS) Practice Exam 2026 Edition
Q6. Which distribution is used for modeling binary outcomes?
A. Normal distribution
B. Bernoulli distribution
C. Poisson distribution
D. Exponential distribution
Answer: B
Explanation: Bernoulli distribution models events with two outcomes (success/failure).
Q7. p-value < 0.05 indicates:
A. Strong evidence against null hypothesis
B. Accept null hypothesis
C. Weak evidence against alternative hypothesis
D. Model is invalid
Answer: A
Explanation: p-value less than significance level suggests rejecting the null.
Q8. Central Limit Theorem states that:
A. Population mean equals sample mean
B. Sample means approximate normal distribution as sample size increases
C. Standard deviation is zero
D. All variables are independent
Answer: B
Explanation: Sample means of a large enough sample are normally distributed.
Q9. Correlation coefficient of -0.8 indicates:
A. Strong positive correlation
B. Strong negative correlation
C. No correlation
D. Weak correlation
Answer: B
Explanation: Negative correlation shows an inverse relationship.
Q10. Which metric is most appropriate for imbalanced classification?
A. Accuracy
B. Precision, Recall, F1-score
C. Mean Squared Error
D. R-squared
Answer: B
Explanation: Precision and recall handle class imbalance better than accuracy.
Machine Learning – Supervised Learning
,Senior Data Scientist (SDS) Practice Exam 2026 Edition
Q11. Linear regression assumes:
A. Linear relationship between independent and dependent variables
B. No multicollinearity
C. Homoscedasticity
D. All of the above
Answer: D
Explanation: These assumptions are critical for linear regression validity.
Q12. Logistic regression outputs:
A. Continuous values
B. Probabilities of classes
C. Clusters
D. None of the above
Answer: B
Explanation: Logistic regression predicts class probabilities.
Q13. Decision trees are prone to:
A. Overfitting
B. Underfitting
C. Linear regression bias
D. None of the above
Answer: A
Explanation: Trees can perfectly fit training data, causing high variance.
Q14. Random Forest improves Decision Trees by:
A. Bootstrapping samples
B. Feature randomness
C. Averaging predictions
D. All of the above
Answer: D
Explanation: Random Forest reduces overfitting and increases robustness.
Q15. Gradient Boosting differs from Random Forest by:
A. Sequentially improving weak learners
B. Using parallel trees
C. Not using loss functions
D. Being unsupervised
Answer: A
Explanation: Boosting iteratively corrects errors of prior models.
Q16. K-Means clustering requires:
A. Number of clusters
, Senior Data Scientist (SDS) Practice Exam 2026 Edition
B. Distance metric
C. Iterative optimization
D. All of the above
Answer: D
Explanation: K-Means clusters data by minimizing within-cluster distances.
Q17. Hierarchical clustering produces:
A. Dendrogram
B. Probability outputs
C. Regression coefficients
D. Confusion matrix
Answer: A
Explanation: Hierarchical clustering represents nested clusters as a dendrogram.
Q18. PCA is used for:
A. Dimensionality reduction
B. Regression
C. Classification
D. Clustering evaluation
Answer: A
Explanation: PCA transforms data to principal components with maximum variance.
Q19. Silhouette score measures:
A. Cluster cohesion and separation
B. Model accuracy
C. Regression error
D. Confusion in classification
Answer: A
Explanation: Silhouette evaluates how well clusters are separated and cohesive.
Q20. DBSCAN is effective for:
A. Arbitrary-shaped clusters
B. Large outliers
C. Density-based clustering
D. All of the above
Answer: D
Explanation: DBSCAN groups dense regions and identifies outliers.
Q21. Which Python library is used for data manipulation?
A. NumPy
B. pandas
C. matplotlib
Q1. Which of the following best describes supervised learning?
A. Learning patterns without labeled outputs
B. Learning patterns with labeled outputs
C. Only clustering data
D. Only dimensionality reduction
Answer: B
Explanation: Supervised learning trains models on labeled data to predict outcomes.
Q2. Unsupervised learning is primarily used for:
A. Regression
B. Classification
C. Clustering and pattern discovery
D. Time series forecasting
Answer: C
Explanation: Unsupervised learning finds hidden structures without labeled outputs.
Q3. Which of the following is a continuous variable?
A. Number of cars
B. Temperature in Celsius
C. Gender
D. Country
Answer: B
Explanation: Continuous variables can take any real value within a range.
Q4. In data science, bias refers to:
A. Systematic error in predictions
B. Random noise
C. Sample size
D. Data visualization choice
Answer: A
Explanation: Bias causes a model to consistently underperform in certain areas.
Q5. Variance in a model measures:
A. How far predictions are from the mean
B. Consistency of predictions across datasets
C. Both bias and error
D. Only model accuracy
Answer: B
Explanation: High variance indicates overfitting to training data.
,Senior Data Scientist (SDS) Practice Exam 2026 Edition
Q6. Which distribution is used for modeling binary outcomes?
A. Normal distribution
B. Bernoulli distribution
C. Poisson distribution
D. Exponential distribution
Answer: B
Explanation: Bernoulli distribution models events with two outcomes (success/failure).
Q7. p-value < 0.05 indicates:
A. Strong evidence against null hypothesis
B. Accept null hypothesis
C. Weak evidence against alternative hypothesis
D. Model is invalid
Answer: A
Explanation: p-value less than significance level suggests rejecting the null.
Q8. Central Limit Theorem states that:
A. Population mean equals sample mean
B. Sample means approximate normal distribution as sample size increases
C. Standard deviation is zero
D. All variables are independent
Answer: B
Explanation: Sample means of a large enough sample are normally distributed.
Q9. Correlation coefficient of -0.8 indicates:
A. Strong positive correlation
B. Strong negative correlation
C. No correlation
D. Weak correlation
Answer: B
Explanation: Negative correlation shows an inverse relationship.
Q10. Which metric is most appropriate for imbalanced classification?
A. Accuracy
B. Precision, Recall, F1-score
C. Mean Squared Error
D. R-squared
Answer: B
Explanation: Precision and recall handle class imbalance better than accuracy.
Machine Learning – Supervised Learning
,Senior Data Scientist (SDS) Practice Exam 2026 Edition
Q11. Linear regression assumes:
A. Linear relationship between independent and dependent variables
B. No multicollinearity
C. Homoscedasticity
D. All of the above
Answer: D
Explanation: These assumptions are critical for linear regression validity.
Q12. Logistic regression outputs:
A. Continuous values
B. Probabilities of classes
C. Clusters
D. None of the above
Answer: B
Explanation: Logistic regression predicts class probabilities.
Q13. Decision trees are prone to:
A. Overfitting
B. Underfitting
C. Linear regression bias
D. None of the above
Answer: A
Explanation: Trees can perfectly fit training data, causing high variance.
Q14. Random Forest improves Decision Trees by:
A. Bootstrapping samples
B. Feature randomness
C. Averaging predictions
D. All of the above
Answer: D
Explanation: Random Forest reduces overfitting and increases robustness.
Q15. Gradient Boosting differs from Random Forest by:
A. Sequentially improving weak learners
B. Using parallel trees
C. Not using loss functions
D. Being unsupervised
Answer: A
Explanation: Boosting iteratively corrects errors of prior models.
Q16. K-Means clustering requires:
A. Number of clusters
, Senior Data Scientist (SDS) Practice Exam 2026 Edition
B. Distance metric
C. Iterative optimization
D. All of the above
Answer: D
Explanation: K-Means clusters data by minimizing within-cluster distances.
Q17. Hierarchical clustering produces:
A. Dendrogram
B. Probability outputs
C. Regression coefficients
D. Confusion matrix
Answer: A
Explanation: Hierarchical clustering represents nested clusters as a dendrogram.
Q18. PCA is used for:
A. Dimensionality reduction
B. Regression
C. Classification
D. Clustering evaluation
Answer: A
Explanation: PCA transforms data to principal components with maximum variance.
Q19. Silhouette score measures:
A. Cluster cohesion and separation
B. Model accuracy
C. Regression error
D. Confusion in classification
Answer: A
Explanation: Silhouette evaluates how well clusters are separated and cohesive.
Q20. DBSCAN is effective for:
A. Arbitrary-shaped clusters
B. Large outliers
C. Density-based clustering
D. All of the above
Answer: D
Explanation: DBSCAN groups dense regions and identifies outliers.
Q21. Which Python library is used for data manipulation?
A. NumPy
B. pandas
C. matplotlib
