EZE, CLETUS NKECHUKWU (2026) DEVELOPMENT OF A STUDENT ATTENDANCE AND REPORTING SYSTEM FOR GODFREY OKOYE UNIVERSITY USING MACHINE LEARNING. Other thesis, Godfrey Okoye University, Enugu Nigeria.
|
Text
FINAL_YEAR_PROJECT.pdf Download (1MB) | Preview |
Abstract
Manual attendance management at Godfrey Okoye University is operationally inefficient, suffers from proxy fraud, and generates data that arrives too late for early intervention with poor attendees, thus impeding enforcement the National Universities Commission (NUC) minimum attendance policy of 75% for all students to qualify to take examination. This study aimed to design, develop, and deploy SmartSync, a cloud-native Smart Attendance and Reporting System, to replace the existing paper-based attendance procedure with a secure, automated, and real-time digital alternative for Godfrey Okoye University. The study was carried out with the use of the Incremental Model Development within the Systems Development Life Cycle (SDLC) framework. The Backend Application Programming Interface (API) was built with the Go programming language, while the frontend was developed using Next.js and React. To combat attendance fraud, SmartSync restricts scans to one device per registered student, through the usage of device binding, and employ rotating Quick Response (QR) codes that change every five seconds and cryptographically-signed via Hash-based Message Authentication Code (HMAC-SHA256) using Secure Hash Algorithm 256-bit (SHA-256) to mitigate the usage of proxy based on screenshots, the system has a 3-state machine (Absent, Signed- In, Completed) to authenticate both scan in and scan out. A non-blocking Asynchronous Processing Queue system which leverages Redis Streams with retry and dead-letter mechanisms was used to mitigate attendance-based traffic spikes. Also, a Server-Sent event (SSE) infrastructure was put in place to stream students’ rosters to lectures with almost instantaneous display. Finally, a machine learning classifier (Extreme Gradient Boosting (XGBoost)) was trained on a weekly attendance feature set and is deployed in the system to identify students who are at high risk of failing courses so as to allow for early intervention. Docker Compose is used to provision, deploy and containerize the whole stack and is exposed to the internet through over Hypertext Transfer Protocol Secure (HTTPS) using Caddy as a reverse proxy with automatic certificate updates. SmartSync has passed all tests and is capable of successfully thwarting proxy attempts on mock-run tests, correctly imposing the required states, and concurrently handling a massive number of students with minimal delays and absolutely no data loss. On the deployed machine, it was proven that our machine Learning (ML) model’s accuracy in detecting risky attendees has an 89% precision, 91% recall,90% F1-score, the harmonic mean of precision and recall, and an Area Under the Receiver Operating Characteristic Curve (AUC-ROC) of 0.94 score on validation data, all of which exceeded our initial expectations. SmartSync demonstrates that a combination of layered security controls, asynchronous queue-based processing, real-time streaming, and machine learning classification can effectively eliminate the core weaknesses of manual attendance management, offering a scalable, fraud-resistant, and data-driven attendance solution for Godfrey Okoye University and similar institutions.
| Item Type: | Thesis (Other) |
|---|---|
| Subjects: | Q Science > Q Science (General) |
| Divisions: | Faculty of Engineering, Science and Mathematics > School of Electronics and Computer Science |
| Depositing User: | Cynthia Ugwuoti |
| Date Deposited: | 22 Jul 2026 11:02 |
| Last Modified: | 22 Jul 2026 11:02 |
| URI: | http://eprints.gouni.edu.ng/id/eprint/5990 |
Actions (login required)
 |
View Item |
