🧠 Intelligence Module – Overview

The Intelligence Module is the analytical heart of the Student Retention Platform. It fuses academic results, engagement scores, financial data, and sentiment signals into predictive insights that help universities intervene before a student decides to leave.

Vision: Turn scattered campus data into crystal‑clear forecasts and prescriptive recommendations.

---

🚀 Key Capabilities

Capability	Description
Risk Scoring	ML models output a 0–1 probability of attrition for each student.
What‑If Analysis	Simulate GPA improvements, housing changes, or aid increases.
Cohort Comparison	Identify patterns by program, year, or demographic segment.
Intervention ROI	Quantify which actions deliver the highest retention lift.
Alert Triage	Auto‑prioritize cases based on severity and time‑sensitivity.

---

🏗️ Architecture Snapshot

1. Data Lake → Raw SIS, LMS, CRM, Financial‑Aid tables.
2. Feature Store → Normalized, time‑indexed student features.
3. Model Layer → XGBoost + SHAP interpretability & rules engine.
4. API Gateway → /risk-score, /recommendations, /kpi endpoints.
5. Visualization → React + D3 dashboards in Docusaurus.

---

📊 KPIs Tracked

Metric	Target	Frequency
Rolling 12‑mo Retention Rate	+3 pp	Monthly
Average Risk Score (≤0.25 good)	↓	Weekly
Intervention Success Rate	↑	Term‑end
Model AUC	≥0.85	Quarterly

---

👤 Personas Served

• Provost & Deans – strategic dashboards, trend lines.
• Data Analysts – drill‑down notebooks & model metrics.
• Advisors – “next‑best action” cards for each advisee.
• Developers – secure REST/GraphQL interfaces.

---

🔮 Roadmap Highlights

Phase	Milestone	ETA
MVP	Baseline Risk Model + CSV ingest	Q3‑25
v1	Real‑time Scoring & SHAP explanations	Q1‑26
v2	Prescriptive AI (intervention ranking)	Q3‑26

---

📚 Further Reading

• White‑paper: Predictive Analytics for Higher‑Ed (PDF)
• JIRA Epic: RET‑INT‑201
• Data Dictionary: /docs/intelligence/data-model.mdx