Hype vs. Reality: Integrated Sentiment Analysis and Machine Learning for Pre-Release Movie Revenue Prediction

This study presents a novel approach to predicting movie revenue by combining sentiment analysis from YouTube trailer comments with machine learning. Given the lack of tailored datasets for this purpose, we created a comprehensive dataset spanning 2013 to 2024. This dataset integrates audience sentiment from trailers and key pre-release features like cast, genre, and budget. Using the BERT model for sentiment analysis, we achieved an impressive accuracy on the ACLIMDB dataset, effectively capturing early audience reactions. These insights, combined with machine learning models, enhance revenue predictions by incorporating both subjective audience feedback and objective movie features. By addressing gaps in existing research, this work offers a reliable framework for informed decision-making in the film industry. All code and data are available in the project repository here [1]