Data-Driven Assessment of BSIT Internship Performance: A Predictive Analytics Approach in a Hybrid Training Framework

Abstract

Purpose - The aim is to analyze the performance of BSIT interns using a hybrid internship model by creating predictive analytics tools that measure job readiness. This study integrates on-site and supplemental evaluation data to find the most significant factors that predict employability results.

Methods - The dataset includes evaluation scores from the Center for Linkages and Placement (CLP) and the College of Computing and Information Sciences (CCIS), encompassing both technical and soft skills. Some of the steps done before processing are encoding, normalizing, and feature engineering. Four supervised machine learning models—Logistic Regression, Random Forest, SVM, and KNN—were trained using an 80/20 split and validated with 5-fold cross-validation. Model performance was measured using accuracy, precision, recall, F1-score, and ROC-AUC.

Results – The Random Forest classifier was the most accurate and easiest to understand of all the models tested. The key predictors were certifications earned, attendance at seminars, and personality attributes, including problem-solving and professionalism. Both technical and developmental activities have a substantial impact on internship performance scores, as shown in the results.

Implications –Predictive analytics can be utilized as a strategic instrument in curriculum development and other academic decision-making by identifying early indicators for job readiness. Institutions can use these findings to implement interventions, align internship programs with industry standards, refine hybrid training frameworks, and promote data-driven performance monitoring.

Conclusion – The results show that predictive analytics is a strong way to judge the performance of BSIT interns. Combining formal evaluations from different parts gives a fuller picture of job preparedness, which helps with evidence-based internship management.

Recommendations – Institutions should integrate predictive models into internship dashboards for real-time monitoring and encourage students to engage in certifications and seminars. Future studies may incorporate qualitative feedback and peer assessments to enhance model accuracy and depth.