Muhammad Rizki Wahyudie, - (2025) OPTIMASI MODEL NEURAL COLLABORATIVE FILTERING UNTUK MENINGKATKAN AKURASI PREDIKSI DALAM REKOMENDASI FILM BERDASARKAN RATING: Studi Kasus: MovieLens. S1 thesis, Universitas Pendidikan Indonesia.
Abstract
Pesatnya perkembangan internet menyebabkan lonjakan data yang memicu kelebihan informasi di industri hiburan, khususnya film. Rekomendasi Film menjadi solusi untuk membantu pengguna menemukan konten yang relevan. Neural Collaborative Filtering (NCF), sebagai pendekatan deep learning, menawarkan keunggulan dalam memodelkan interaksi pengguna dan item dibandingkan metode tradisional. Penelitian ini bertujuan untuk mengimplementasikan dan mengoptimalkan model NCF dalam sistem rekomendasi film menggunakan dataset MovieLens 100K, serta menganalisis dampak hyperparameter tuning, termasuk variasi learning rate, batch size, dan jenis optimizer, terhadap kinerja model untuk meningkatkan akurasi prediksi rekomendasi. Evaluasi dilakukan menggunakan Root Mean Square Error (RMSE) dan Mean Absolute Error (MAE). Hasil penelitian menunjukkan bahwa penggunaan optimizer Nadam dengan learning rate 0,00002 dan batch size 256 memberikan tingkat error yang rendah, dengan RMSE sebesar 0,8697 dan MAE sebesar 0,6678. Selain itu, nilai loss train sebesar 0,4236 dan loss validation sebesar 0,7950 menunjukkan konsistensi model dalam proses pelatihan dan validasi. Hyperparameter tuning terbukti efektif dalam meningkatkan performa model, menjadikannya lebih adaptif terhadap karakteristik data. Indikator penilaian menunjukkan bahwa model ini memiliki kinerja yang sangat baik dalam meningkatkan akurasi rekomendasi film. Temuan ini memberikan wawasan praktis bagi pengembang sistem rekomendasi dalam memilih parameter optimal untuk meningkatkan akurasi dan pengalaman pengguna. --------- The rapid development of the internet has led to a surge in data, causing information overload in the entertainment industry, particularly in films. Movie recommendations have become a solution to help users find relevant content. Neural Collaborative Filtering (NCF), a deep learning approach, offers advantages in modeling user-item interactions compared to traditional methods. This study aims to implement and optimize the NCF model in a movie recommendation system using the MovieLens 100K dataset, and to analyze the impact of hyperparameter tuning, including variations in learning rate, batch size, and optimizer type, on model performance to improve recommendation accuracy. Evaluation is performed using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). Results show that using the Nadam optimizer with a learning rate of 0.00002 and a batch size of 256 yields low error rates, with an RMSE of 0.8697 and an MAE of 0.6678. Additionally, the training loss of 0.4236 and validation loss of 0.7950 demonstrate consistency in both training and validation processes. Hyperparameter tuning is shown to effectively improve model performance, making it more adaptive to data characteristics. The evaluation indicators show that the model performs excellently in improving recommendation accuracy. These findings provide practical insights for recommendation system developers to choose optimal parameters to enhance accuracy and user experience.
![[img]](http://repository.upi.edu/style/images/fileicons/text.png) |
Text
S_RPL_2001151_Title.pdf Download (569kB) |
|
Text
S_RPL_2001151_Chapter1.pdf Download (52kB) |
|
Text
S_RPL_2001151_Chapter2.pdf Restricted to Staf Perpustakaan Download (822kB) |
|
Text
S_RPL_2001151_Chapter3.pdf Download (531kB) |
|
Text
S_RPL_2001151_Chapter4.pdf Restricted to Staf Perpustakaan Download (761kB) |
|
Text
S_RPL_2001151_Chapter5.pdf Download (39kB) |
|
Text
S_RPL_2001151_Appendix.pdf Restricted to Staf Perpustakaan Download (213kB) |
| Item Type: | Thesis (S1) |
|---|---|
| Uncontrolled Keywords: | Dataset MovieLens, Hyperparameter tuning, Neural Collaborative Filtering, Rekomendasi Film, Root Mean Square Error (RMSE) |
| Subjects: | Q Science > QA Mathematics > QA76 Computer software T Technology > T Technology (General) |
| Divisions: | UPI Kampus cibiru > S1 Rekayasa Perangkaat Lunak |
| Depositing User: | Muhammad Rizki Wahyudie |
| Date Deposited: | 07 Mar 2025 04:07 |
| Last Modified: | 07 Mar 2025 04:07 |
| URI: | http://repository.upi.edu/id/eprint/130168 |
Actions (login required)
 |
View Item |