Sistem Rekomendasi Peminjaman Buku dengan Menggunakan User Based Collaborative Filtering

Abstract

Reading is very useful in increasing understanding and opening up insight into the intellectual worldviews. But there are some problems that are often encountered by readers, especially for students is determining the books they will read next. This case also be found for readers at the Ceria Bangsaku Community Reading Park. The difficulty of readers in determining which books to read is caused by the variety of the books. The solution of the problems is to implement a recommendation system that can provide book recommendations to them. This research applied the user-based collaborative filtering method because this method can provide recommendation results based on similarities between users. The test results showed the average MAE value for 3 neighbors was 0.49. The 5 neighbors scenario of 0.36. With an all neighbors scenario of 0.2. The lowest MAE value using the all neighbors scenario was 0.2, which is more accurate than the other scenarios. In testing the average MAPE value for 3 neighbors was 28.62%. With a scenario of 5 neighbors amounting to 19.26%. The all neighbors scenario of 9.9%. The lowest MAPE value of 9.9% uses the all neighbors scenario with a MAPE range of <10%. So it can be said that the MAPE results with all neighbors have very good prediction.