Year: 2025 | Month: May | Volume: 12 | Issue: 5 | Pages: 321-330
DOI: https://doi.org/10.52403/ijrr.20250535
Trigger Based Smart Attendance Framework with Machine Learning for Predictive Student Performance Analysis
Diptika Ghosh Roy1, Abir Bhattacharjee1, Rakesh Das1, Prabhat Das1,2
1Department of Computer Science and Engineering, School of Engineering and Technology, Adamas University, Kolkata, India
2Department of Information Technology, School of Computing Sciences, The Assam Kaziranga University, Assam, India
Corresponding Author: Prabhat Das
ABSTRACT
In educational institutions, traditional attendance-taking methods such as manual roll calls or sign-in sheets are inefficient and prone to issues like proxy attendance and human error. These limitations compromise data integrity, hinder performance monitoring, and increase the burden on administrative staff. To address these challenges, this paper proposes a Trigger-Based Smart Attendance Framework that integrates biometric fingerprint recognition, IoT microcontroller hardware, and machine learning algorithms for automated, secure, and predictive attendance tracking. The system employs ESP32 microcontrollers and R307S optical fingerprint sensors installed at individual classroom benches, allowing students to log attendance using their unique biometric identifiers. A master ESP32 unit coordinates communication among the fingerprint scanners and transmits attendance data to a centralized server via Wi-Fi. Upon successful fingerprint authentication, a relay module triggers a secondary device (slave ESP32), adding an extra layer of verification and control.
On the software side, a web-based platform built with React.js offers intuitive, role-based interfaces for students, teachers, and administrators. It includes real-time dashboards for attendance viewing, performance monitoring, and system control. The backend, developed using Flask (a Python web framework), enables RESTful API communication and manages data storage in a MongoDB NoSQL database. Beyond data collection, the system integrates Random Forest and XGBoost machine learning models to analyze attendance patterns and predict student performance. This predictive capability supports early identification of at-risk students, enabling timely academic intervention. Overall, the proposed framework enhances operational efficiency, transparency, and data-driven decision-making, marking a significant step toward digital transformation in academic administration.
Keywords: Smart Attendance System, Biometric Authentication, IoT in Education, Machine Learning, Student Performance Prediction
[PDF Full Text]