Collaborative Filtering for Steam Games Recommendation

This research explores collaborative filtering with Singular Value Decomposition (SVD) and Pearson correlation to provide game recommendations on the Steam digital distribution service dataset. Collaborative Filtering aims to leverage user preferences to identify similar gaming patterns. This can be done through SVD and Pearson correlation algorithms. SVD reduces the dimensionality of the game rating matrix, while Pearson correlation measures the similarity between users and games. The proposed method generates personalized recommendations based on user preferences and opinions of similar gamers. Evaluation metrics include Mean Average Error (MAE) and Root Mean Square Deviation (RMSE). Results demonstrate the effectiveness of CF with SVD and Pearson correlation in delivering relevant and personalized game suggestions on Steam based dataset.