A HYBRID MODEL FOR PEDICTON OF STUDENT PERFORMANCE AND CAREER

ABSTRACT vii Academic underperformance and poor career alignment remain major challenges facing Nigerian students, often resulting in low employability, graduate unemployment, and job dissatisfaction. Traditional evaluation methods focus mainly on grades and test scores, while career counseling is limited and mostly influenced by parental or societal pressure. This project therefore designed and developed a Student Performance and Career Prediction System that integrates machine learning techniques to provide accurate academic predictions and personalized career guidance. A hybrid approach was employed, combining Logistic Regression and Decision Tree models for academic performance prediction, while Random Forest was applied for career recommendation. Data obtained from Kaggle was preprocessed through cleaning, transformation, and feature selection to ensure reliability before training and testing the models. Evaluation showed that the Logistic Regression model achieved an accuracy of 88% in predicting student performance categories, while the Random Forest career model recorded an outstanding accuracy of 99% in generating career recommendations. The system was deployed as a web-based application using Streamlit, making it interactive, user-friendly, and accessible to both students and academic counselors. By considering academic records alongside non-academic attributes such as teamwork, communication skills, problem-solving, and analytical ability, the system provides holistic and evidence-based guidance. The results demonstrate that the system can effectively identify at-risk students, suggest academic interventions, and align student strengths with suitable career paths. The study contributes to knowledge by introducing a two-stage hybrid framework that links academic prediction with career guidance, offering a practical solution to reduce career mismatch, improve student performance, and enhance employability in Nigeria.