Penerapan Locality Preserving Projection Untuk Meningkatkan Kualitas Klasterisasi Data Film

Lailiyah, Nabiilah (2025) Penerapan Locality Preserving Projection Untuk Meningkatkan Kualitas Klasterisasi Data Film. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002211008-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Pertumbuhan industri film yang pesat di era digital mendorong kebutuhan akan sistem rekomendasi yang efisien untuk membantu penonton menemukan film sesuai preferensi mereka. Salah satu pendekatan yang umum digunakan adalah klasterisasi film berdasarkan fitur konten. Namun, tingginya dimensi data film sering kali menghambat kualitas klasterisasi. Penelitian ini bertujuan untuk menerapkan metode Locality Preserving Projection (LPP), serta menganalisis hasil klasterisasi data film berdasarkan kemiripan konten dan preferensi penonton. Data yang digunakan berasal dari Movielens, dengan representasi fitur genre menggunakan One-Hot Encoding, tag dengan Word2Vec, dan rating berdasarkan statistik rating. Hasil reduksi dimensi untuk gabungan tiga fitur menghasilkan lima komponen utama dengan trustworthiness = 0.941, yang kemudian dilakukan klasterisasi dengan tiga algoritma : K-Means, Gaussian Mixture Models (GMM), dan Balanced Iterative Reducing and Clustering using Hierarchies (BIRCH0 . Evaluasi hasil klasterisasi dilakukan menggunakan Silhouette Score, Davies-Bouldin Index (DBI), dan Calinski-Harabasz Index (CHI), serta visualisasi 2D dan 3D. Hasil menunjukkan bahwa LPP meningkatkan kualitas klasterisasi, di mana K-Means memberikan hasil terbaik. Klaster hasil reduksi LPP menunjukkan karakteristik isi yang terdefinisi, di mana kombinasi fitur genre, tag, dan rating mampu membentuk kelompok film yang tidak hanya menggambarkan kesamaan tematik, tetapi juga preferensi dan persepsi penonton. Dengan demikian, LPP terbukti efektif untuk meningkatkan kualitas klasterisasi data film serta menghasilkan kelompok data film yang lebih informatif dan bermakna. Hasil penelitian ini dapat dimanfaatkan dalam pengembangan sistem rekomendasi dan analisis konten digital.
===================================================================================================================================
The rapid growth of the film industry in the digital era has driven the need for efficient recommendation systems to help viewers discover films that match their preferences. One commonly used approach is film clustering based on content features. However, the high dimensionality of film data often hinders clustering quality. This research aims to implement the Locality Preserving Projection (LPP) method and analyze film data clustering results based on content similarity and viewer preferences. The data used is derived from MovieLens, with genre feature representation using One-Hot Encoding, tags with Word2Vec, and ratings based on rating statistics. The dimensionality reduction results for the combination of three features yielded five principal components with trustworthiness = 0.941, which were then clustered using three algorithms: K-Means, Gaussian Mixture Models (GMM), and Balanced Iterative Reducing and Clustering using Hierarchies (BIRCH). Evaluation of clustering results was conducted using Silhouette Score, Davies-Bouldin Index (DBI), and Calinski-Harabasz Index (CHI), as well as 2D and 3D visualization. The results show that LPP improves clustering quality, with K-Means providing the best results. Clusters resulting from LPP reduction demonstrate well-defined content characteristics, where the combination of genre, tag, and rating features is capable of forming film groups that not only represent thematic similarity but also viewer preferences and perceptions. Thus, LPP proves effective in improving film data clustering quality and generating more informative and meaningful film data groups. The results of this research can be utilized in the development of recommendation systems and digital content analysis.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	BIRCH, Evaluasi Internal, GMM, K-Means, Klasterisasi Film, Locality Preserving Projection, Reduksi Dimensi, BIRCH, Dimensionality Reduction, GMM, Internal Evaluation, Movie Clustering, K-Means, Locality Preserving Projection
Subjects:	N Fine Arts > N Visual arts (General) For photography, see TR > N7433.8 Virtual reality Q Science Q Science > QA Mathematics Q Science > QA Mathematics > QA278.55 Cluster analysis Q Science > QA Mathematics > QA278.5 Principal components analysis. Factor analysis. Correspondence analysis (Statistics)
Divisions:	Faculty of Mathematics and Science > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Nabiilah Lailiyah
Date Deposited:	01 Aug 2025 07:32
Last Modified:	01 Aug 2025 07:33
URI:	http://repository.its.ac.id/id/eprint/125895

Actions (login required)

View Item