ML Problem Framing Quiz - Competitive Exams GCP_ML_Engineer

Free practice quiz: ML Problem Framing (Competitive Exams GCP_ML_Engineer). 50 multiple-choice questions with answers and explanations. Play online, no sign-up needed.

This Competitive Exams GCP_ML_Engineer practice quiz has 50 multiple-choice questions on ML Problem Framing. Try the 5 sample questions below (tap an option to check yourself), then play the full quiz online. No sign-up needed.

Sample questions

1. When framing a machine learning problem, what is the primary purpose of defining a clear objective?

   1. To determine the cloud provider
   2. To guide selection of data, features, and evaluation metrics
   3. To choose the programming language
   4. To ensure the problem is solvable with ML
   Show answer

   Answer: B. To guide selection of data, features, and evaluation metrics

2. What does the term 'label' refer to in supervised learning?

   1. The input features
   2. The model parameters
   3. The output or target variable
   4. The evaluation metric
   Show answer

   Answer: C. The output or target variable

3. Which of the following is an example of a semi-supervised learning problem?

   1. Using only unlabeled data
   2. Using a small labeled dataset and a large unlabeled dataset
   3. Using reinforcement learning with rewards
   4. Using only labeled data
   Show answer

   Answer: B. Using a small labeled dataset and a large unlabeled dataset

4. What term describes the situation where the offline metric improves but the business KPI does not, because the offline metric is insufficiently tied to the actual goal?

   1. Data drift
   2. Metric misalignment
   3. Concept drift
   4. Overfitting
   Show answer

   Answer: B. Metric misalignment

5. When the labels for a task are themselves noisy or generated by heuristics, how should the problem framing account for this?

   1. Treat the label noise as a modeling challenge and consider probabilistic labels or noise-robust loss functions
   2. Switch to an unsupervised approach regardless of the task
   3. Increase model complexity until the noise fits
   4. Ignore label quality since models are robust
   Show answer

   Answer: A. Treat the label noise as a modeling challenge and consider probabilistic labels or noise-robust loss functions

More questions in this quiz

Answers and explanations are in the full quiz.

1. Which of the following is a key characteristic of a supervised learning problem?
2. In an ML problem where the goal is to predict a continuous value, what type of problem is it?
3. What is the first step in framing an ML problem according to best practices?
4. When the target variable is categorical with two possible outcomes, the problem is typically framed as:
5. Which of the following best describes an unsupervised learning problem?
6. What is a common pitfall when framing an ML problem?
7. In ML problem framing, why is it important to consider data availability and quality?
8. Which type of ML problem involves learning from interactions with an environment to maximize cumulative reward?
9. When framing a problem where you need to detect fraudulent transactions among millions of legitimate ones, what is a key consideration?
10. Which of the following is an example of a multi-class classification problem?
11. In ML problem framing, what is the purpose of defining an evaluation metric?
12. Which of the following problems is best suited for anomaly detection?
13. What is the difference between a classification and a regression problem?
14. In framing an ML problem, what does 'feature engineering' involve?
15. Which ML problem type is most appropriate for grouping news articles by topic without predefined categories?
16. What is a key question to ask when determining if an ML solution is appropriate for a business problem?
17. In the context of ML problem framing, what is a 'proxy metric'?
18. When framing an ML problem, why is it important to consider the cost of different types of errors?
19. What is the primary goal of exploratory data analysis (EDA) in ML problem framing?
20. In ML problem framing, what does 'label leakage' refer to?
21. Which of the following best describes a 'batch' vs 'online' prediction problem?
22. When framing an ML problem, what is the significance of the 'baseline' model?
23. In CRISP-DM, which phase immediately follows 'Business Understanding' and is concerned with inventorying and describing the available data?
24. A data science team is choosing a success metric for a model that prioritizes recall-based clinical screening. Which metric family should they select?
25. Which framework is commonly used to document ML product requirements and success metrics before development begins?
26. A governance board rejects an ML proposal because individual decisions must be fully explainable and appealable. Which characteristic makes the problem unsuitable for ML?
27. Which ML learning approach is specifically designed for environments where obtaining labels is expensive but vast unlabeled data is available?
28. Which method for splitting time-dependent data prevents lookahead bias during model evaluation?
29. Which type of machine learning problem is best characterized by delayed, sparse, and indirect feedback signals?
30. How does a hit rate metric differ from a business KPI when framing an ML problem?
31. In problem framing, why should the ML objective be aligned with the business objective?
32. Which approach for handling imbalanced classification problems reframes the task as predicting the probability of the minority class?
33. What does it mean when a problem is described as 'human-in-the-loop' in an ML system?
34. Framing an ML problem as a 'ranking' task rather than a classification task is most appropriate when:
35. A model is evaluated with a micro-averaged metric versus a macro-averaged metric. What does the choice primarily affect?
36. Which statement best reflects a data-centric framing decision rather than a model-centric one?
37. Which technique translates a future event prediction problem into a standard supervised classification framing?
38. What is the main reason to define a 'holding out' test split during problem framing rather than after model development?
39. Which of the following best characterizes a 'weak supervision' approach in problem framing?
40. In defining ML problem scope, what does the 'minimum viable model' concept suggest?
41. Why might a problem with perfectly deterministic outcomes (e.g., arithmetic) be a poor fit for ML framing?
42. Which framing consideration is most relevant for a decision problem where different prediction errors carry asymmetric financial impact?
43. What is the role of a 'problem statement' document in ML project framing?
44. When the cost of acquiring labels is prohibitive, which framing alternative is most appropriate if the goal is to rank items by relevance?
45. How does framing a task as 'prediction' differ from framing it as 'causal inference'?

More quizzes

• Model Deployment & Monitoring▶ Play
• Vertex AI & Pipelines▶ Play
• Data Engineering for ML▶ Play

Browse all GCP_ML_Engineer quizzes

More to play