Deteksi Komunitas Pada Jaringan Peserta Event Menggunakan Algoritma Leiden

Dika, Felicia Farah (2026) Deteksi Komunitas Pada Jaringan Peserta Event Menggunakan Algoritma Leiden. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002221099-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (17MB) | Request a copy

Abstract

Analisis pola partisipasi peserta pada berbagai event diperlukan untuk memahami keterkaitan antarpeserta serta mengidentifikasi kelompok peserta yang memiliki kecenderungan partisipasi yang serupa. Namun, hubungan antarpeserta tidak dapat diamati secara langsung dari data partisipasi sehingga diperlukan pendekatan analisis jaringan untuk mengungkap struktur komunitas yang terbentuk. Penelitian ini bertujuan mendeteksi komunitas pada jaringan peserta event menggunakan Algoritma Leiden. Dataset yang digunakan merupakan Community Events dari Kaggle yang terdiri atas 49.060 data partisipasi, 3.750 peserta, 40 event, dan 8 kategori ProgramType. Jaringan peserta dibangun menggunakan pendekatan co-occurrence network, yaitu hubungan antarpeserta dibentuk berdasarkan kesamaan kehadiran pada event yang sama. Selanjutnya, bobot hubungan dinormalisasi menggunakan metode association strength untuk mengurangi bias akibat perbedaan tingkat partisipasi peserta sehingga kekuatan hubungan antarpeserta menjadi lebih proporsional. Struktur komunitas kemudian dideteksi menggunakan Algoritma Leiden dan dievaluasi berdasarkan metrik modularity, conductance, dan density. Hasil analisis menunjukkan bahwa jaringan berhasil dikelompokkan menjadi delapan komunitas dengan nilai modularity sebesar 0,868639. Nilai conductance yang berada pada rentang 0,003517–0,009220 mengindikasikan bahwa setiap komunitas memiliki pemisahan yang baik terhadap komunitas lain, sedangkan nilai density pada rentang 0,999447–1,000000 menunjukkan tingkat keterhubungan internal komunitas yang sangat tinggi. Analisis karakteristik komunitas memperlihatkan bahwa setiap komunitas memiliki ProgramType dominan yang berbeda, yaitu Arts, Technology, Education, Culture, Wellness, Fitness, Environment, dan Support. Dengan demikian, kombinasi pendekatan co-occurrence network, normalisasi association strength, dan Algoritma Leiden terbukti mampu menghasilkan struktur komunitas yang representatif untuk menganalisis pola partisipasi peserta berdasarkan kesamaan kehadiran pada berbagai event.
=====================================================================================================================================
Analyzing participant participation patterns across various events is essential for understanding the relationships among participants and identifying groups that exhibit similar participation tendencies. However, these relationships cannot be directly observed from participation data, requiring a network analysis approach to uncover the underlying community structure. This study aims to detect communities within an event participant network using the Leiden Algorithm. The dataset used is the Community Events dataset from Kaggle, consisting of 49,060 participation records, 3,750 participants, 40 events, and 8 ProgramType categories. The participant network was constructed using a cooccurrence network approach, where relationships between participants were established based on their co-attendance at the same events. Subsequently, edge weights were normalized using the association strength method to reduce bias caused by differences in participants’ activity levels, resulting in more proportional relationship strengths between participants. The community structure was then detected using the Leiden Algorithm and evaluated based on the modularity, conductance, and density metrics. The results show that the network was successfully partitioned into eight communities with a modularity value of 0.868639. The conductance values, ranging from 0.003517 to 0.009220, indicate well-separated communities, while the density values, ranging from 0.999447 to 1.000000, demonstrate a very high level of internal connectivity within each community. Furthermore, the community characteristic analysis reveals that each community is dominated by a distinct ProgramType, namely Arts, Technology, Education, Culture, Wellness, Fitness, Environment, and Support. These findings demonstrate that the combination of the co-occurrence network approach, association strength normalization, and the Leiden Algorithm is effective in producing a representative community structure for analyzing participant participation patterns based on shared event attendance.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Deteksi komunitas, Algoritma Leiden, analisis jaringan, co-occurrence network, association strength, Community detection, Leiden Algorithm, network analysis
Subjects:	Q Science Q Science > QA Mathematics
Divisions:	Faculty of Mathematics and Science > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Felicia Farah Dika
Date Deposited:	29 Jul 2026 06:27
Last Modified:	29 Jul 2026 06:27
URI:	http://repository.its.ac.id/id/eprint/139471

Actions (login required)

View Item