Final project description
Students will act as a fraud analytics team for CDE, an EBT/SNAP payments processor working with state agencies. Their task is to analyze EBT‑style transaction data synthetically, identify fraud and abuse patterns at the cardholder, merchant, and terminal levels, and propose actionable detection rules, analytics, and controls that CDE can implement while balancing fraud reduction with equitable access to benefits.
Requirements (15-page APA paper)
1. Format and length
Minimum 15 pages of main text (excluding title page, references, and any appendices).
APA 7th edition student paper format.
Double‑spaced, 1‑inch margins, 12‑point Times New Roman (or similar).
Page numbers in the header; standard paragraph indentation.
2. Required sections
You can list these as mandatory headings in the assignment sheet:
Introduction and context
Explain EBT/SNAP programs and CDE’s role as a payments processor.
State the purpose of the project and key research questions (e.g., “What fraud patterns can a processor detect, and how should CDE respond?”).
Background and literature/industry review
Summarize major types of EBT/SNAP fraud (card compromise, trafficking, merchant fraud, misuse).
Briefly cover relevant regulations, state/federal program guidance, and typical fraud‑control practices.
Use and cite official or scholarly sources.
Data and methodology
Describe the dataset used (fields, entities: card, merchant, terminal, transactions; real or synthetic).
Explain feature engineering (e.g., velocity metrics, geography distance, peer benchmarks, terminal risk flags).
Describe the analytic approach: rule‑based indicators, statistical analysis, and/or any basic models used (e.g., decision tree, clustering).
Fraud patterns and findings
Present the key patterns you identified at:
Cardholder level (velocity, geography, amount patterns, compromise indicators).
Merchant level (volume anomalies, concentration, repeated amounts, high-risk profiles).
Terminal/device level (skimming indicators, high decline/PIN retry rates).
Include at least:
One table summarizing key metrics or patterns.
One figure (diagram, simple chart, or dashboard mock‑up) to illustrate your analytics.
Detection rules and controls for CDE
Propose specific rules or scoring logic CDE could implement (e.g., thresholds, risk scores, watchlists).
Discuss how these would be operationalized: alerts, case management, hot‑card/hot‑merchant lists, card controls, and messaging to state agencies.
Issues, limitations, and ethics
Discuss data and model limitations, including false positives and missing ground truth.
Address fairness and equity concerns (risk of over‑targeting certain merchants or communities; impact on legitimate users).
Reflect on how to balance fraud prevention with timely access to benefits.
Conclusions and recommendations
Summarize main insights.
Provide 2–3 prioritized recommendations for CDE and state partners (e.g., immediate rules, medium‑term analytics roadmap, technology upgrades).
Suggest opportunities for further work (e.g., real‑time monitoring, advanced ML, cross‑state data sharing).
References
Minimum of 8–10 credible sources (government reports, academic articles, industry white papers, standards).
All sources cited in APA 7th edition format; consistent in-text citations.
3. Evaluation criteria (you can adapt to a rubric)
Problem understanding and framing (15%)
Clear explanation of EBT, CDE’s role, and the fraud problem; well‑articulated objectives.
Use of sources and context (15%)
Quality, relevance, and integration of literature/industry sources; correct APA citation.
Data and methodology (20%)
Appropriate and clearly explained data design, feature engineering, and analytic methods.
Fraud pattern analysis (20%)
Depth and clarity of identified patterns at card, merchant, and terminal levels; use of tables/figures.
Recommendations and critical reflection (20%)
Practicality and specificity of rules/controls; thoughtful discussion of limitations, ethics, and tradeoffs.
Writing quality and formatting (10%)
Organization, clarity, grammar, adherence to APA style, and the 15-page requirement.