COMPARATIVE EVALUATION OF SOME CLASSIFICATION ALGORITHMS FOR SENTIMENT ANALYSIS OF COMMENTS ON YOUTUBE EDUCATIONAL VIDEOS

The virtual learning environment also known as the online learning space has expanded over time with the development of information technology. Individuals now have the opportunity to benefit from the teaching-learning process and air their views. YouTube, an example of public online space on the web has an added advantage because it allows engagement between learners and teachers. Its videos are also accompanied by social information which captures information about learners' sentiments and experiences. Sentiment analysis helps to identify the emotional tone behind a body of text, classify and determine the motives behind opinionated content on the platform. This work aims to compare Random Forest (RF), Support Vector Machine (SVM), eXtreme Gradient Boosting (XGB) and Light Gradient Boosting Machines (LGBM) to see how well they can classify the sentiments of learners written in English. We compared the performance of the models with TF-IDF and BoW feature extraction methods, concatenated features, balanced and imbalanced datasets and two sampling strategies. Furthermore, we evaluated these models with standard performance metrics. SVM had the highest accuracy of 0.9017