Evaluating Machine Learning Approaches for Personalized Movie Recommendations: A Comprehensive Analysis

Platforms and movie theatres provide a large range of movies that need to be filtered to each user's tastes. For this objective, recommender systems are a useful tool. This research presents a novel hybrid recommender system for personalized movie suggestions, which integrates content-based methods with collaborative filtering. This study develops a personalized movie recommendation system utilizing the MovieLens 1M dataset, comprising user ratings for a diverse set of movies. The research data undergoes separation into training segments that constitute 80% of the total sample while testing comprises 20% of the data. The evaluation framework incorporates multiple metrics which include F1-Score together with Precision and Root Mean Square Error (RMSE) as well as Mean Absolute Error (MAE) alongside Recall. A Multilayer Perceptron (MLP) model is employed for movie recommendations and compared to a Deep Neural Network (DNN). The outcomes show that a MLP model outperforms the DNN, achieving a lower RMSE of 0.99 and an MAE of 0.80, in contrast to the DNN’s RMSE of 1.011 and MAE of 73.5. The training and validation loss trajectories show continuous progress and maintain minimal avoidable patterns. Future work will refine the model accuracy by performing hyperparameter tweaking and developing advanced feature extraction processes together with implementing distinct deep learning models for enhanced recommendation capability and user satisfaction.