# Session 04 Assignment: Fraud Detection

## Goal

Study fraud detection as both a classification problem and an anomaly detection problem.

## Recommended Demos

- `DomainUseCaseDemos\CreditCards\CreditCardFraud`
- `DomainUseCaseDemos\CreditCards\CreditCardTxnFraud`
- `DomainUseCaseDemos\CreditCards\CreditCardFraudOutlier001`

## Student Tasks

1. Generate synthetic fraud data.
2. Run one classification workflow and one outlier workflow.
3. Compare precision, recall, and operational trade-offs.
4. Explain why fraud detection should not be judged only by accuracy.

## Required Outputs

- confusion matrix comparison
- metric summary
- one-page operational interpretation

## Extension Ideas

- tune contamination or threshold settings
- simulate higher fraud prevalence
- suggest alert triage rules for operations teams