CAP CAP Methodology Selection & Analytics Problem Framing 1 — Questions and Answers
Question 1: Which step in translating a business problem into an analytics problem involves defining the unit of analysis?
- Identifying the target variable
- Specifying the grain of the data (Correct answer)
- Selecting the model type
- Choosing evaluation metrics
Correct answer: Specifying the grain of the data
Specifying the grain defines what each record represents, which is the unit of analysis for the analytics problem.
Question 2: A retail company wants to reduce customer churn. Which analytics approach is MOST appropriate as a first step?
- Descriptive analytics to profile churned customers (Correct answer)
- Prescriptive analytics to set retention budgets
- Real-time streaming analytics
- Graph analytics for social networks
Correct answer: Descriptive analytics to profile churned customers
Descriptive analytics on churned customers first reveals patterns that then inform predictive or prescriptive approaches.
Question 3: When selecting between supervised and unsupervised learning, the key distinguishing factor is:
- The size of the dataset
- Whether labeled outcome data is available (Correct answer)
- The number of input features
- The computing resources available
Correct answer: Whether labeled outcome data is available
Supervised learning requires labeled outcomes, while unsupervised learning discovers structure without predefined labels.
Question 4: Which CAP domain involves converting a business objective into a measurable analytics goal?
- Data management
- Analytics problem framing (Correct answer)
- Model deployment
- Risk mitigation
Correct answer: Analytics problem framing
Analytics problem framing translates vague business goals into specific, measurable analytics objectives with defined success criteria.
Question 5: A company needs to segment customers with no prior labeling scheme. Which methodology is MOST suitable?
- Logistic regression
- K-means clustering (Correct answer)
- Random forest classification
- Time-series forecasting
Correct answer: K-means clustering
K-means clustering is an unsupervised method ideal for discovering natural groupings when no predefined labels exist.
Question 6: Which factor is MOST important when deciding between a parametric and non-parametric model?
- The analyst's personal preference
- Whether the data meets distributional assumptions (Correct answer)
- The color coding of the dashboard
- The vendor of the analytics platform
Correct answer: Whether the data meets distributional assumptions
Parametric models require specific distributional assumptions (e.g., normality), so violating those assumptions favors non-parametric alternatives.
CAP CAP Methodology Selection & Analytics Problem Framing 1
Which step in translating a business problem into an analytics problem involves defining the unit of analysis?