Skip to content

AI_Powered_Student_Success_PRD.md

Latest commit

 

History

History
136 lines (103 loc) · 6.21 KB

AI_Powered_Student_Success_PRD.md

File metadata and controls

136 lines (103 loc) · 6.21 KB

Product Requirements Document (PRD)

Project Title: AI-Powered Student Success Analytics
Hackathon Theme: Data, AI/ML, and Visualization for Higher Education

Institutions Involved:

  • Kentucky Community & Technical College System (KCTCS)
  • Bishop State Community College
  • University of Akron

1. Overview

This project aims to create a unified AI-powered analytics and visualization platform that improves student readiness, retention, and institutional decision-making across diverse higher-education systems.

The solution will connect existing institutional data pipelines (PDP, AR files, and internal data warehouses) to produce real-time dashboards, predictive insights, and natural language query capabilities.

2. Problem Statement

Institution Primary Challenges
KCTCS - Underprepared students not visible in weekly enrollment reports.
- Limited insight into readiness metrics.
- Need to integrate PDP/AR data for proactive interventions.
Bishop State - Fragmented data access (single PDP admin).
- Difficulty generating reports without IT intervention.
- Need for AI-assisted dashboards and chat interfaces for faculty and leadership.
University of Akron - PDP dashboards underutilized.
- Data is siloed (SharePoint, PowerBI).
- Opportunity to use 8 years of PDP data for predictive modeling of student success and retention.

3. Goals and Objectives

  • Data Integration: Automate ingestion and harmonization of PDP, AR, and institutional datasets.
  • AI Insights: Empower non-technical users to ask natural language questions and generate dashboards instantly.
  • Predictive Analytics: Forecast retention rates and student outcomes based on historical and real-time data.
  • Accessibility: Democratize access to institutional data for advisors, faculty, and leadership.
  • Impact: Improve student retention by 5–15% and enable early intervention strategies.

4. Target Users

Role Needs
Advisors Identify at-risk students, access real-time dashboards, and personalize interventions.
Faculty View course combinations linked to student drop/failure/withdraw rates.
Institutional Research Teams Automate PDP validation, track submission errors, and monitor readiness metrics.
Leadership / Deans Export insights and visualizations for board presentations and grant proposals.

5. Key Features

5.1 AI-Powered Dashboard

  • Natural Language Query Interface: Ask, e.g., "Show first-year students not passing gateway courses."
  • Instant Visualization: Auto-generate PowerBI-style charts or graphs.
  • Weekly Refresh: Integrate with institutional data warehouses for near-real-time updates.

5.2 Predictive Analytics Module

  • Retention Forecasting: Predict which cohorts are at risk.
  • Readiness Scoring: Quantify preparedness using DFWI and gateway completion data.
  • Trend Identification: Analyze course combinations correlated with success/failure.

5.3 Data Integration Layer

  • Connectors: PDP, AR, CSV, SharePoint, Oracle (PeopleSoft), AWS Data Warehouse.
  • Validation Script: Python-based PDP file checker to ensure format compliance before submission.
  • Storage: Unified schema in Supabase (PostgreSQL) for hackathon MVP.

5.4 Visualization and Reporting

  • Role-Based Dashboards: Advisors, Leadership, IR teams.
  • Export Formats: CSV, PDF, or embedded dashboards for presentations.

KPI Examples:

  • Retention % by cohort
  • Readiness Index by major
  • Gateway course completion rate
  • DFWI ratio by course sequence

6. Architecture Overview

Data Flow:

Institutional Systems (PDP, AR, Oracle, CSV)
         ↓
Data Ingestion Layer (ETL/Conduit/Custom Python Pipelines)
         ↓
Supabase (PostgreSQL + APIs)
         ↓
AI Query Engine (LangChain + OpenAI API)
         ↓
Visualization Layer (Streamlit/PowerBI/React Dashboard)

Proposed Tools:

  • Backend: Supabase (data access + API endpoints + Auth)
  • Data Pipeline: Conduit (Meroxa) or Airbyte
  • Database: Supabase (Postgres)
  • Frontend: React + PowerBI Embedded / Streamlit
  • ML/AI: Scikit-Learn / Hugging Face Transformers / LangChain
  • Hosting: AWS / Fly.io

7. Success Metrics

Metric Description Target
Retention Prediction Accuracy ML model accuracy predicting at-risk students ≥ 85%
Dashboard Latency Time to visualize query results ≤ 5 seconds
Data Refresh Frequency Automatic weekly refresh 1x/week
User Adoption Number of unique advisors/faculty using dashboards +25% per semester
ROI Impact Estimated revenue preserved from improved retention ≥ $500K/year per institution

8. Roadmap (24-Hour Sprint)

Phase Duration Deliverables
Phase 1 – Discovery & Setup 3 hours Data mapping, Supabase setup, schema design, MVP scope
Phase 2 – Data Pipeline & Dashboard 8 hours ETL pipeline, Supabase integration, basic dashboard prototype
Phase 3 – AI Layer & Analytics 8 hours Natural language querying, basic predictive model, visualizations
Phase 4 – Polish & Demo Prep 5 hours Bug fixes, presentation deck, live demo rehearsal, documentation

9. Potential Extensions

  • Real-Time Alerts: Weekly email alerts to advisors highlighting high-risk students.
  • Chatbot Interface: "Ask your data" using a chat widget integrated into PowerBI.
  • Course Optimization Tool: Suggest ideal course sequences to reduce DFWI rates.
  • Benchmark Dashboard: Compare institution metrics against state or national averages.

10. Risks and Mitigations

Risk Mitigation
Data privacy (FERPA/PII) Anonymize data before AI processing; use secure storage.
Inconsistent PDP formats Use automated validation and schema mapping.
Limited hackathon time Focus MVP on one institution dataset and scale post-demo.
Adoption resistance Include faculty in pilot feedback loop; show time saved.

11. Deliverables for Hackathon Presentation

  • Live dashboard demo
  • Short AI query walkthrough
  • Sample predictive report
  • Visual architecture diagram
  • One-page ROI summary per institution