Applied Data Science Lab - end-to-end tweet sentiment analysis pipeline using RoBERTa (WorldQuant University)
Built an end-to-end NLP pipeline to classify scraped tweets into sentiment categories and power a live dashboard with real-time predictions. Prepared and used model-ready text inputs from web-scraped data for downstream sentiment classification and analytics. Benchmarked and integrated RoBERTa-based predictions into visualization workflows for interactive review. • Generated sentiment labels (positive/neutral/negative) from tweet text • Integrated CardiffNLP RoBERTa model outputs into a Streamlit app • Enabled real-time text prediction and time-series trend visualization • Supported exploratory analysis via word clouds and live sentiment dashboards