CDSP 8 Quiz

1. What is a key goal of communicating results to stakeholders in a data science project?

A: To present complex data models without explanation
B: To translate findings into business insights that demonstrate value
C: To provide unfiltered raw data for stakeholders to analyze
D: To focus solely on the technical aspects of the project

2. Why is explainability important in data science models?

A: It makes models more efficient
B: It ensures that the models are complex enough for all scenarios
C: It helps stakeholders trust the model's decision-making process
D: It reduces the accuracy of the model

3. What is SHAP used for in data science?

A: To explain local model behavior
B: To provide global feature importance in a model
C: To reduce the size of datasets
D: To improve the model's performance

4. What is an advantage of using visualizations when communicating results to non-technical audiences?

A: They increase the complexity of the data
B: They simplify the presentation of data patterns
C: They eliminate the need for labels and legends
D: They make the data harder to understand

5. What is a common issue when using neural networks for explainability?

A: Neural networks are too simple to need explanation
B: Neural networks often act as "black boxes"
C: Neural networks cannot handle large datasets
D: Neural networks provide too much transparency

6. What is one purpose of a dashboard in data science projects?

A: To store all data collected from the project
B: To give stakeholders a quick visual overview of project progress
C: To hide complex metrics from stakeholders
D: To remove KPIs from the project

7. What is one benefit of Explainable AI (XAI)?

A: It reduces the need for data in model training
B: It promotes transparency and trust in AI decision-making
C: It completely eliminates the "black box" issue in all models
D: It guarantees higher accuracy for all AI models

8. What does the term “local interpretability” refer to in model explainability?

A: Understanding how a model performs across an entire dataset
B: Explaining a specific prediction or a small subset of data
C: Visualizing global trends in model behavior
D: Creating an entirely new model

9. What type of models are commonly considered "black boxes"?

A: Decision trees
B: Logistic regression
C: Neural networks
D: Linear regression

10. What is one common use of cumulative gains charts in data science?

A: To visualize the compactness of clusters
B: To show the proportion of positive responses from a model
C: To explain the accuracy of a neural network
D: To simplify feature selection

11. What does the principle of transparency in data science emphasize?

A: Making predictions more accurate
B: Ensuring that the inner workings of a process are visible
C: Hiding irrelevant data points
D: Reducing the complexity of machine learning models

12. What is a recommended guideline for using visuals in presentations?

A: Avoid labeling axes to keep the presentation simple
B: Focus on using complex chart types for all audiences
C: Label axes and chart titles clearly and use legends where needed
D: Only use line charts regardless of data type

13. What is one key feature of web apps when demonstrating data science models?

A: Web apps provide static, unchangeable results
B: Web apps allow dynamic input and adjustments to the model
C: Web apps are easier to develop than presentations
D: Web apps cannot integrate visualizations

14. What is the primary purpose of a production pipeline in a data science project?

A: To permanently store model results
B: To automate the sequence of tasks for deploying models
C: To eliminate the need for human oversight
D: To increase the complexity of the data

15. What is a key feature of the ETL process in data pipelines?

A: It focuses on eliminating all data from the project
B: It prepares and organizes data before analysis
C: It only works with unstructured data
D: It limits data transformation

16. What is the purpose of using Docker in a data science project?

A: To permanently store data
B: To create virtual containers for isolating and managing project components
C: To increase the size of datasets
D: To reduce the need for cloud storage

17. What does the concept of "model drift" refer to in data science?

A: A model’s tendency to become more accurate over time
B: Changes in the underlying patterns of data that affect model performance
C: The gradual reduction in the size of the dataset
D: The static nature of a deployed model

18. What is a key advantage of using Fivetran in data pipelines?

A: It only works with structured data
B: It automates the entire ETL process with minimal impact on source systems
C: It eliminates the need for data governance
D: It provides a free, unlimited version for all users

19. Why is securing data pipelines important for businesses?

A: To speed up the data transformation process
B: To protect sensitive data from unauthorized access and potential attacks
C: To eliminate the need for model maintenance
D: To reduce data storage costs

20. What is one challenge of managing big data in a data pipeline?

A: Ensuring data is permanently deleted after processing
B: Handling the high volume, variety, and velocity of data efficiently
C: Reducing the quality of data to make it manageable
D: Eliminating the need for data storage