The rapid development of internet users over the last decade has led to an increase in the use of electronic commerce (e-commerce). The existence of a recommender system influences the success of e-commerce. Collaborative Filtering (CF) is one of the most frequently used recommender system methods. However, in real cases, sparsity problems generally occur. This is generally caused because only a small number of users give ratings to items. In this study, we propose the combination of clustering and dimension reduction methods on the Amazon Review Data to overcome the sparsity problem. The clustering method with K-Means is used to group users based on item preferences. Meanwhile, we used Singular Value Decomposition (SVD) for dimension reduction to improve the performance of the recommender system in sparse data. The results show that the combination of SVD and K-Means is successful in predicting ratings with an RMSE value of less than 2, significant performance increase compared to previous study. The use of SVD is proven to be able to overcome sparsity, with a decrease in RMSE of 9.372%.