Evaluating Machine Learning Models for Sentiment Analysis of YouTube Comments and Creating an Accessible Web Application for Comment Analysis

The popularity of the YouTube platform for content creation and its rapid accessibility to users is growing significantly. YouTube is eager to receive feedback on the material they have produced and made available to the public. Most research in this field has focused on analyzing static data rather than real-time streaming data. This study provides comprehensive investigation of machine learning-based sentiment analysis of live YouTube comments. This research paper proposes the development of an effective sentiment analysis model with user friendly web application that can accu- rately classify live YouTube comments into three groups: neutral, negative, and positive. The machine learning models, like extreme gradient boosting, random forest, logistic regression, adaptive boosting, gradient boosted decision tree, and bagging classifier along with natural language processing techniques have been used to implement the proposed system. In this research work, we used the Kaggle dataset to train the machine learning models and evaluated the performance. performance of these models using performance metrics like accuracy, precision, recall, and f1 score. The experimental work showed that the bagging classifier with 95.8 % accuracy performed well as compared to others. Finally, the bag- ging classifier was deployed to create a user-friendly web interface to make the developed system an effective YouTube comment analyzer for live and online video. Hence, by using this system, YouTube content creators can quickly analyze the impact of their videos on society.