Perbandingan Algoritma Cosine Similarity dan Euclidean Distance pada Sistem Rekomendasi Film dengan Metode Item-Based Collaborative Filtering

The current recommendation system is increasingly needed along with the increasing number of movies available, especially in digital media. Item-Based Collaborative Filtering is one of the many methods in the recommendation system. The Item-Based Collaborative Filtering method determines recommended movies based on similarities to other movies based on other movies that have been rated. In a series of processes in Item-Based Collaborative Filtering, there is one step with an algorithm to determine the similarity or similarity between items. This study compares two algorithms to determine the similarity between items. The algorithms compared are Cosine Similarity and Euclidean Distance. Both algorithms were applied and tested on the movie recommendation system using the Item-Based Collaborative Filtering method on MovieLens movie rating data. Comparisons are made by calculating the Mean Absolute Error and Root Mean Square Error values ​​to evaluate the accuracy of each algorithm used. In an experiment using the Cosine Similarity algorithm, it produces an accuracy value with a Mean Absolute Error of 2.21 and a Root Mean Square Error value of 2.51. While the experiment using the Euclidean Distance algorithm produces an accuracy value with a Mean Absolute Error of 2.24 and a Root Mean Square Error value of 2.55. From the results of the research that has been done, the Cosine Similarity algorithm has a better accuracy rate than the Euclidean Distance algorithm. This can be seen from the Mean Absolute Error and Root Mean Square Error values ​​in the Cosine Similarity algorithm which have a smaller value than the Euclidean Distance algorithm.