<![CDATA[The number of existing anime is increasingly varied and more in line with the increasing number of enthusiasts. The surge in anime series among anime enthusiasts has become an obstacle to finding anime that matches their taste. This underlies the writer to create an anime recommendation system using User-Based Collaborative Filtering method. The research process consisted of several stages, namely data collection from the Kaggle website with 3 pieces of data uploaded, namely in the .csv format. Determination of users who have a correlation, using the Spearman Rank Correlation Coefficient method. Calculation of predictions using a weighted sum algorithm. The final stage is the implementation of the recommendations and evaluation of the recommendation system used to calculate the level of collaborative filtering using the Mean Absolute Error (MAE).. This research has output in the form of a website which has several components, namely Home Page, Login-Register, Search, Recommend, Result Page, Single View and Rating. Testing on the system uses MAE calculations which are carried out on 50 users with the most rating history. The results from the test show that the percentage of error obtained is 15.8% and the prediction accuracy results obtained are 84.12%. The smallest MAE value of the 50 profiles is 0.894933222 by Archaeon and the highest MAE value is 3.572438553 by Krunchyman.]]>
Copyrights © 2023
