About EduRisk

Revolutionizing student success through artificial intelligence and IoT

The Challenge

Every year, thousands of students struggle academically without early intervention. Traditional monitoring systems identify issues too late.

35% of at-risk students are identified only after failing

Reactive approaches lead to dropout rates of 15-20% in engineering programs.

Faculty spend 10+ hours/week on manual monitoring

Time that could be spent on meaningful interventions.

Data silos prevent holistic view

Attendance, grades, and engagement exist in separate systems.

78% of dropouts

can be predicted by the end of the first semester with ML models

Source: Educational Data Mining Research 2024

Our Solution

Early Detection

Machine learning models analyze academic patterns to flag at-risk students 8-12 weeks before traditional methods.

IoT Integration

RFID attendance, library access, and lab usage provide real-time engagement data to the ML pipeline.

Actionable Insights

Personalized recommendations for students and alerts for faculty enable timely interventions.

The Technology Behind It

Machine Learning Pipeline

  • Random Forest & XGBoost classifiers for risk prediction
  • LSTM networks for temporal academic patterns
  • Feature engineering from 50+ academic variables
  • Model accuracy: 89.7% on test data

IoT Infrastructure

  • Raspberry Pi-based RFID scanners at entry points
  • QR code attendance system via mobile app
  • Real-time data synchronization with cloud
  • Edge computing for instant processing

Analytics Dashboard

  • Real-time risk heatmaps by department/semester
  • Individual student profiles with risk factors
  • Intervention tracking and effectiveness metrics
  • Exportable reports for academic committees

Recommendation Engine

  • Personalized study resources based on weak areas
  • Automated tutor/alumni mentoring suggestions
  • Course load optimization recommendations
  • Peer study group formation algorithm

Expected Impact

35%

Reduction in academic probation cases

50%

Faster intervention time

89%

Prediction accuracy

24/7

Real-time monitoring