<![CDATA[Recommender system is widely implemented in various fields. Collaborative Filtering is one of the most used recommender system paradigms because it is easy to use. K-means clustering algorithm is widely use in Collaborative Filtering. This algorithm can predict the item rating that will be given by a user. Rating can be predicted by calculating the average rating of the item. The clustering performance of this algorithm is low because this algorithm selects initial centroid randomly. This causes high errors in the item rating prediction. To obtain lower error, we propose dimensionality reduction with Singular Value Decomposition (SVD). SVD is able to factorize the clustering result data and reduce dimensionality of the data. Dimensionality reduction with SVD can be carried out by removing non-dominant characteristics of the data. This study uses the result of factorization to calculate the similarity between clusters. The value of similarity between clusters is used to predict the rating of an item that will be given by a cluster. The experimental results show that the combined method of K-means and SVD can produces RMSE up to 8.936% lower than the K-means method.]]>
Copyrights © 2022
