This recommendation system was designed to predict the ratings of movies unseen by users according to the genres and tags associated with the movies they have liked or disliked. The results produced will return the top ten movies with the highest predicted ratings.
Predictions are a result of three models. The first model feeds in the genres from the movies that the like and disliked, and also includes the genres of the movie in question into a neural network/deep learning model and outputs a predicted rating. The second model is similar to the first model except it uses the tags associated with the movies that have been watched by the user and the tags associated with the movie in question to predict a rating. The third model stacks the two predicted ratings from the first two models and predicts a final rating using linear regression.