Students' Experiences on Covid-19: An Automatic Classification Using Machine Learning-Based Sentiment Analysis

The COVID-19 epidemic that struck the world in 2020 caused severe disruptions to the economy and society. The digitization of instruction has brought about a significant change in the realm of higher education. With this, Cagayan State University – Sanchez Mira introduces innovative techniques to combat the negative effects of pandemic to students' educational, social, family, personal and health journey. The main purpose of this study is to classify the experiences of students of CSU-Sanchez Mira during COVID-19 outbreak. In this experiment, machine learning was done using RapidMiner. The first level of categorization involved labeling statements as having good or negative experiences. The second level of segmentation entailed grouping experiences into categories based on how they related to personal, family, education, social, and health experiences. Statements that are unfavorable outnumber those that are favorable. Since the focus of this study focuses on the viewpoints of students, the bulk of statements were on experiences related to school. The f-score and accuracy increase when the aforementioned traits are added to the classifier, indicating that the features enhance the classifier. The majority of ratings were average, according to the results, which may have been a result of noise and insufficient data. Nevertheless, every accuracy score that was reported during the test was more than 60%, indicating a relatively accurate classification. The Naive Bayes approach has the most accurate classification outcomes. The 80.65% accuracy of Naive Bayes classification throughout the experiment is promising, aligning well with recent outcomes in opinion mining research in recent years.