A Categorical Transformer with a Data Science Approach for Recommendation Systems Based on Collaborative Filtering

Recommender systems help predict what customers might like, such as movies, restaurants, or products. Collaborative filtering, a crucial part of these systems, faces challenges when dealing with user, item, and rating data. Traditional machine learning struggles with this data because user and item data are categorical. To solve this, we propose a method that transforms the original data into new variables, making it more suitable for advanced machine learning and deep learning techniques. This approach enhances prediction quality and opens doors for innovative data processing methods in collaborative filtering.