Analyzed Australian Steam user and game metadata of ~800,000 unique user-game pairs over 30000+ games
Generated models in Python to recommend games to users with Logistic Regression, Support Vector Machines, Overall Game Popularity, and Jaccard/Cosine Similarity, and Pearson Correlation
Final Logistic model with log transformed variables and Jaccard Similarity recommended with 90.25% accuracy
Final Report can be found in "Steam Game Recommendations for Australian Users - Recommender Systems"