A Random Forest Approach for Real-Time Sentiment Analysis of Twitter Data

Abstract

Twitter sentiment analysis has become pivotal in understanding public emotions and trends. This study presents a comprehensive framework leveraging the Random Forest algorithm to classify sentiments expressed in tweets into positive, negative, and neutral categories. The dataset comprises 75,681 entries, including sentiment labels and text, with 74,995 unique tweet texts. Extensive preprocessing and feature extraction techniques, including tokenization and lemmatization, ensure data readiness for machine learning. The deployment of a Streamlit based application enhances usability by providing real-time analysis and visualization. Experimental results demonstrate high classification accuracy, emphasizing the effectiveness of the Random Forest model for sentiment analysis tasks. The integration of machine learning and web technologies highlights the potential of this system in real-world applications such as brand management, political analysis, and customer feedback monitoring.

Keywords: Sentiment Analysis, Random Forest, Machine Learning