Prepare to pass the CAP-Expert across all seven INFORMS domains

A sponsor's draft problem statement reads: "Improve operational efficiency in the claims department." What is the most important deficiency to resolve before proceeding?

Tap an answer to reveal the correct answer, rationale and source.

• A. It never names a software platform or vendor stack for delivery
• B. It omits a line-item project budget in the draft write-up.
• C. It lacks a measurable objective and an explicit success criterion.
• D. It never names a modeling technique or algorithm family.

Before acquiring data, an analyst maps required fields back to the hypothesized drivers from framing. The primary purpose is:

Tap an answer to reveal the correct answer, rationale and source.

• A. To maximize the number of columns collected regardless of decision relevance, acquisition cost, or hypothesized driver coverage from framing.
• B. To set the project deadline before any needs prioritization occurs in framing.
• C. To choose the fitting algorithm before clarifying which inputs matter for the decision.
• D. To prioritize acquisition by relevance to the decision, so effort targets attributes that matter rather than everything available.

The analytics problem is a supervised binary classification with a need for per-case explanations to regulators. Which candidate-method shortlist best fits *at the identification stage*?

Tap an answer to reveal the correct answer, rationale and source.

• A. Prioritize only deep neural networks for maximum accuracy at identification, treating per-case regulatory explanations as a post-hoc step that does not constrain the initial shortlist.
• B. Shortlist time-series ARIMA as the primary candidate family because temporal smoothing can stand in for binary classification when regulators request case-level explanations.
• C. Favor unsupervised clustering at identification, inferring segments that regulators can treat as per-case explanations for individual approval decisions.
• D. Interpretable classifiers (e.g., logistic regression, decision trees, or explainable/monotonic models) as primary candidates, given the explanation constraint.