Text-Based Sentiment Analysis of Students' Satisfaction: An Input to Strategic Plan

Abstract

Sentiment analysis is the study of people's sentiments as expressed in written language. To ensure continuous improvement in higher education institutions (HEIs), it is critical to identify the sentiments of each stakeholder. City College of Calamba (CCC), as one of the recognized HEIs in the Philippines, utilized the traditional method to collect the sentiments of its stakeholders, which is the use of a suggestion box. Despite the pandemic, the suggestion box remains the primary platform for collecting the sentiments of its stakeholders and clients. In this paper, the most commonly used sentiment classification algorithms were Multinomial Nave Bayes (MNB) and Linear Support Vector Machine (LSVM). The main objective of the study is to analyzethe sentiments of CCC stakeholders retrieved from the suggestion box, and the results may serve as an input to the strategic plan of the institution to plan the necessary improvements. The method employed in the study has the following phases: data preprocessing, classification, validation, and evaluation. In terms of data classification, the most commonly used algorithms for sentiment classification were evaluated to identify the optimum algorithm for the sentiment analysis of the stakeholders and clients. The results revealed that the LSVM algorithm is the optimal algorithm for analyzing text sentiments. The results obtained using the LSVM algorithm may be used as input to the strengths and weaknesses analysis, which is vital to the strategic plan of the institution.