Comparative analysis of machine learning approaches for sentiment analysis of students' online learning feedback during covid-19

Abstract

Sentiment analysis aids in determining if a person's feelings are neutral, negative, or positive. Many machine learning and deep learning algorithms exist for assessing people's attitudes on various social media networks. Many researchers focused on students' emotional identification. The purpose of this paper is to analyze the sentiments of academic students regarding the online class experience conducted during the COVID-19 pandemic situation. For this work, the Term Frequency-Inverse Document Frequency (TF-IDF) model is used for the feature extraction and comparison of eight machine learning models were tested for the classification, such as Support Vector Classifier, Multinomial Naïve Bayes, Decision Tree, K-Nearest-Neighbors (KNN), Random Forest, AdaBoost Classifier, Bagging Classifier, Extreme Gradient Boosting Classifier (XGB) and F-Score, accuracy, precision, and Recall are the performance criteria examined. With a test accuracy of 0.97 and precision of 1.0, Multinomial Naive Bayes achieves the highest accurate model.