ASA exams

PA

Predictive Analytics

5 discussions

Study strategy

PA: Practise command words with a structured answer plan

“Calculate,” “interpret,” “compare” and “recommend” call for different outputs. A calculation can support an interpretation, but the final number is not an interpretation by itself. Practise reading the task before composing the answer.

Worked example or practice scenario

For “compare two validation approaches,” name both approaches, give a relevant difference, connect it to this dataset and state a consequence. “Approach A is better” omits the comparison and its reason. For “recommend,” make a choice and justify it within the stated constraints.

Try this next

Use publicly released official practice material under realistic time limits, then check for unanswered subparts. Keep an error log for missing implications, unsupported claims and irrelevant paragraphs. This is writing practice, not a prediction of grading rules or a guaranteed pass strategy.

Reading sources

Official syllabus and resources

ActNet editorial guide · October 1, 2026 · Original illustrative examples.

0 repliesOct 1

Explanation

PA: A comparison needs a benchmark and a reason for the metric

Declaring a candidate model “better” requires a defined baseline and criterion. Compare performance on the same held-out observations. Also consider interpretability, stability and deployment constraints; these are reasons to discuss a trade-off, not excuses to ignore poor fit.

Worked example or practice scenario

A tree model reduces validation error by 2% relative to a GLM but changes materially with a small training perturbation. A recommendation should identify both the improvement and the instability, then propose a check or mitigation before implementation.

Try this next

Write a compact comparison: target and metric, benchmark result, candidate result, material limitation and recommended next step. Do not rely on training fit alone. Show why the metric reflects the task instead of choosing whichever statistic flatters your preferred model.

Reading sources

Official syllabus and resources

ActNet editorial guide · October 1, 2026 · Original illustrative examples.

0 repliesOct 1

Explanation

PA: Evaluate calibration as well as ranking

A model can order risks well while systematically overstating or understating their expected outcomes. Ranking and calibration answer different questions. Select diagnostics consistent with the target, exposure and business use.

Worked example or practice scenario

Model A predicts twice the observed claim cost in every broad segment but ranks segments correctly. A ranking statistic may look good while a portfolio pricing estimate is severely overstated. Compare aggregate predicted and observed amounts and inspect segment-level calibration.

Try this next

Construct a table with exposure, observed total, predicted total and the ratio for each segment. Distinguish random fluctuation in a small segment from a stable pattern in a large one. Avoid comparing unweighted averages when the intended result is exposure-weighted.

Reading sources

Official syllabus and resources

ActNet editorial guide · October 1, 2026 · Original illustrative examples.

0 repliesOct 1

Explanation

PA: Leakage can survive an apparently strong validation score

Leakage occurs when a model uses information unavailable at prediction time or when preprocessing allows held-out data to influence the fit. A random split does not fix a variable that was created after the outcome.

Worked example or practice scenario

A claim-cost model uses a case-reserve value recorded after an investigation. If the prediction is intended at policy inception, that field is unavailable. Removing it can worsen measured performance while making the model relevant to its actual use.

Try this next

For every feature, record its creation date and availability at the decision time. Fit imputation and feature selection within the training process. Explain how a proposed time split or exclusion addresses the specific leakage mechanism rather than claiming it removes every bias.

Reading sources

Official syllabus and resources

ActNet editorial guide · October 1, 2026 · Original illustrative examples.

0 repliesOct 1

Explanation

PA: Answer the decision, not just the model output

A useful predictive-analytics answer connects an observation to the requested business decision. State what the result implies, why and what limits the conclusion. More technical detail is not always more responsive to the question.

Worked example or practice scenario

For a retention model, “AUC is .81” is incomplete. A decision-focused explanation might compare capacity-limited outreach groups, identify false-positive cost and say whether the validation sample reflects the next campaign. The selected threshold depends on the intervention, not only AUC.

Try this next

For practice, rewrite one statistical output in three sentences: finding, implication and limitation. Use the audience and command word in the task to decide what to include. Consult current official materials for format; older report-writing threads may describe a different exam version.

Reading sources

Official syllabus and resources

ActNet editorial guide · October 1, 2026 · Original illustrative examples.

0 repliesOct 1