Build a Capacity-Constrained Fraud Review Queue Simulator

A fraud operations manager planning investigator capacity.

Allocate a limited investigator budget across synthetic alerts. Compare expected preventable loss, review effort, customer friction, and uncertainty without executing account restrictions.

Key: alert_id + score_version + planning_period.

Fields: probability, potential_loss, intervention_effectiveness, review_minutes, friction_cost, review_cost.

Expected benefit is conditional on calibration and supplied intervention assumptions. It is not observed savings.

Validate risk probabilities and operational cost inputs.

Calculate expected net review benefit under explicit assumptions.

Optimize the review selection under capacity constraints.

Export queue proposals and sensitivity to costs and calibration.

A queue planner with capacity controls, alternative policies, and expected-versus-observed labels.

A reproducible queue optimization service.

A planning report comparing a simple baseline with the optimized proposal.

Runnable local application, setup commands, and a dependency lockfile.

Versioned input data, data dictionary, acquisition instructions, and rights notes.

Calculation traces, source references, assumption records, and exported review artifacts.

A deterministic baseline and tests for normal, incomplete, contradictory, and failure-recovery cases.

A recorded complete workflow with a reproducible manifest and explicit limitations.

Selected review time stays within capacity.

Negative expected net benefit does not become positive through rounding.

Do not equate risk ranking with measured intervention effectiveness.

Report false-positive workload and expected loss, not accuracy alone.

Identical fraud probability and potential loss can have different net review value when friction costs differ.

The attached starter files are prototype inputs. Expected starter answers are visible. They are not independent holdouts.

Full target: Combine synthetic calibrated risk estimates with fictional loss amounts, review durations, intervention effectiveness, and customer-friction costs. Document all unsupported assumptions.

Complete target contracts and expert review remain pending. Public observations are not historical release vintages.

SciPy, pandas, Plotly

Use an available model appropriate to the task. Record provider, model identifier, version or retrieval date, settings, cost, and latency. Use deterministic code for calculations. Models may extract, retrieve, or explain evidence. They must not invent missing records, override approval rules, or claim professional authority.

All queue actions are simulated. Expected loss reduction is not a proven causal benefit.

No fabricated records, source citations, professional endorsements, or expert-adjudication claims.

No future-data leakage into dated decisions, model fitting, or historical valuation inputs.

No unauthorized payments, account restrictions, regulatory filings, or production changes.

No automated appraisal certification, tax-compliance determination, or unsupported fraud accusation.

No concealed uncertainty or silently invented inputs required for a numerical conclusion.

Correct domain calculations and justified methods with documented assumptions. Weight: 30.

Source lineage, temporal integrity, rights checks, and explicit synthetic-data labels. Weight: 25.

A working application with review, export, error handling, and reproducible state. Weight: 20.

Adversarial cases, uncertainty, missing-data behavior, and sensitivity analysis. Weight: 15.

Evidence-linked conclusions that distinguish facts, assumptions, and unresolved judgments. Weight: 10. Score each dimension from zero to two. Weighted score = sum(weight * score / 2). A failed safety gate blocks acceptance.