In the modern world of online enterprises and e-commerce, there is a tremendous need to engage users by delivering relevant and interesting content. Recommendation systems play a crucial role in providing personalized content recommendations to users, which can improve customer interest, purchase rates, and, ultimately, company profits. This paper aims to categorize and explore different types of recommendation systems, such as collaborative and content-based filtering, and examine various methods and algorithms that can be used for implementing these systems. The accuracy of these algorithms is evaluated using an e-commerce dataset. Experimental results revealed that SVD++ achieved the best and lowest RMSE (Root Mean Square Error) with the parameters 'n epochs': 25, 'reg all': 0.4, and 'lr all': 0.01, while SVD had the second best RMSE with the parameters 'n epochs': 20, 'reg all': 0.2, and 'lr all': 0.005. Comparison between SVD and SVDpp revealed slight differences in RMSE and MAE values, but SVD had significantly shorter Fit Time and Test Time (12 times less) compared to SVDpp. Based on the research and experimental results, SVDpp performed the best in terms of RMSE among the Matrix Factorization Based Algorithms, and KNNWithMeans showed promising results in RMSE among the Collaborative Filtering Algorithms.

Recommendation Systems Collaborative Filtering Content-Based Filtering K-means K- Nearest Neighbor Mean Square Error.