Movie Success Prediction Using Machine Learning Models

Every year, a massive number of films are released. Because of the vast investments made in the film industry, antic-ipating a film's success and minimizing uncertainty early in the film-making process will have a significant influence. Thus, the movie business suffers greatly when a film flops at the box office. Our approach will attempt to forecast a movie's success rate by performing predictive analysis on numerous movie features. This research aims to compare multiple machine-learning techniques for estimating a movie's rating grade among the many movies that exist. Our dataset is extracted from Kaggle which contains data of 7668 movies taken from IMDb (Internet Movie Database) in the form of names of actors and directors, genre, IMDb rating, budget, and others. The implementation results evaluated using various metrics including accuracy, precision, recall, F1 score, and Kappa coefficient show that the Random Forest and XGBoost (Extreme Gradient Boosting) algorithms are effective. The XGBoost approach outperformed all other classifiers tested after being adjusted and employing a bagging ensemble. These findings demonstrate the promise of machine learning techniques for predicting movie success and their practical applications in the film business. The best results were obtained through XGBoost with tuned Bagging with an accuracy rate of 84.68%. While Random Forest after tuning has an accuracy of 82.15%. XGBoost with bagging algorithm shows a great ability to improve prediction performance by combining results from different models.