Development of Machine Learning Models for Sentiment Analysis of University Students' Opinions on NELFUND

Abstract

This study investigated students’ perceptions of the Nigerian Education Loan Fund (NELFUND) at Emmanuel Alayande University of Education, concentrating on issues pertaining to the transparency of loan disbursement and perceived advantages, which influence students’ confidence in the programme. A sentiment analysis methodology was utilised, employing opinion datasets classified into transparency and benefits attributes, with polarity levels predominantly indicating positive sentiments. Data preprocessing encompassed cleaning, tokenisation, part-of-speech tagging, stop-word elimination, stemming, and lemmatisation to enhance data quality. Feature extraction was performed utilising Term Frequency–Inverse Document Frequency (TF-IDF) and count vectorisation methods. Sentiment classification utilised the VADER model, and five machine learning algorithms, Support Vector Machine (SVM), Naïve Bayes (NB), Logistic Regression (LR), Random Forest (RF), and Extreme Gradient Boosting (XGBoost), were assessed based on accuracy, precision, recall, and F1-score metrics. The findings indicated that in terms of transparency-related sentiments, SVM, RF, and XGBoost consistently surpassed NB and LR. Regarding benefits-related sentiments, Random Forest and Support Vector Machine demonstrated optimal performance for both negative and positive polarities. The findings indicate predominantly favourable student perceptions of NELFUND and illustrate the efficacy of advanced machine learning models in analysing policy-related opinions in educational finance.

Keywords: Educational Finance Policy, Extreme Gradient Boosting, Natural Language Processing, Nigerian Education Loan Fund, Sentiment Analysis