Alumni Employability Model with Tracer and Feedback Systems Utilizing Classification and Sentiment Analysis Algorithms

Abstract

Purpose – This study developed and evaluated a Graduate Tracer and Feedback System for Richwell Colleges, Inc. that integrates employability prediction and sentiment analysis in a unified web-based platform to support alumni monitoring, curriculum enhancement, and evidence-based institutional decision-making.

Method – The study used developmental and descriptive research designs guided by the Spiral Model and Knowledge Discovery in Databases process. Logistic Regression, Random Forest, Support Vector Machine, Gradient Boosting, and K-Nearest Neighbors were evaluated using 151 alumni responses for employability prediction. Sentiment analysis models using TF-IDF, machine learning, and deep learning techniques were tested on 83 alumni feedback responses. System acceptability was assessed using the ISO/IEC 25010:2023 Software Product Quality Model with 65 IT experts, administrators, and alumni users.

Results – Logistic Regression obtained the best employability prediction performance with 0.7419 accuracy, 0.8167 precision, 0.6125 recall, 0.6222 F1-score, and 0.5373 Cohen’s Kappa. For sentiment analysis, Gradient Boosting achieved the highest overall performance with 0.706 accuracy, 0.468 precision, 0.497 recall, 0.481 F1-score, and 0.441 Cohen’s Kappa. The system was rated Very Highly Acceptable (WM = 4.87). However, sentiment analysis results should be interpreted cautiously due to the small and imbalanced feedback dataset.

Conclusion – The findings show that integrating employability prediction and sentiment analysis in a graduate tracer platform can support alumni monitoring, curriculum improvement, and analytics-driven decision-making in higher education.

Recommendations – Future studies may use larger multi-institutional datasets, balanced sentiment data, and additional machine learning techniques.

Research Implications – The study contributes to educational data mining by demonstrating an integrated predictive and feedback analysis system for graduate tracer environments.