Analisis Pola Konsumsi Data Trafik pada Operator Seluler dengan Metode Time-Series Clustering

Nova, Muhammad Fachry (2015) Analisis Pola Konsumsi Data Trafik pada Operator Seluler dengan Metode Time-Series Clustering. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6032222061-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (2MB) | Request a copy

Abstract

Seiring dengan berkembangnya teknologi, aktivitas penggunaan internet semakin meningkat setiap tahunnya, tak terkecuali internet dengan jaringan seluler. Operator seluler sebagai penyedia layanan internet perlu melakukan analisis terhadap konsumsi internet di jaringannya, salah satunya adalah data trafik. Operator seluler perlu untuk menganalisis konsumsi data trafik untuk mengetahui potensi dan perilaku penggunanya. Sayangnya, konsumsi data trafik memiliki karakteristik yang selalu berubah berdasarkan waktu tertentu yang bergantung pada perilaku konsumsi pengguna. Karakteristik ini menjadi tantangan bagi operator sehingga perlu metode yang dapat mempelajari pola data trafik dalam domain waktu. Selama ini, operator seluler pada penelitian ini masih menggunakan analisis deskriptif sederhana dalam mengelompokkan potensi data trafik suatu lokasi serta masih terdapat unsur subjektivitas terutama dalam menentukan Point of Interest (POI) pada momen-momen hari libur nasional seperti Idul Fitri, Natal, dan Tahun Baru. Metode ini tentunya membutuhkan proses yang cukup lama dan proses panjang apabila dilakukan karakterisasi setiap lokasi berdasarkan pola konsumsi data trafiknya. Mengingat data trafik merupakan sebuah data time-series, maka diperlukan metode yang dapat mempelajari pola time-series terebut. Pada penelitian kali ini dilakukan analisis konsumsi data trafik dengan memanfaatkan metode time-series clustering untuk mempelajari pola data trafik di 61 Kecamatan pada salah satu operator seluler di Indonesia. Dengan membandingkan beberapa metode clustering yakni: K-Means, PAM, dan Fuzzy C-Means dan memanfaatkan algoritma Dynamic Time Warping (DTW), penelitian ini menghasilkan kluster optimum berdasarkan kemiripan pola dan memiliki karakteristik untuk masing-masing kluster. Metode DTW dengan Fuzzy C-Means memiliki performa paling baik ditinjau dari Silhouette Score dan Davies-Bouldin Index dibandingkan metode lainnya. Dengan metode ini didapatkan tiga kluster optimum yang kemudian dikarakterisasi berdasarkan beberapa parameter yaitu rasio weekend-to-weekday, rasio masa Idul Fitri, rasio masa Natal dan Tahun Baru, rasio Year-to-date. Karakteristik masing-masing kluster menunjukkan bagaimana potensi konsumsi data trafik berdasarkan kondisi waktu tertentu di kecamatan-kecamatan yang telah ditentukan. ==============================================================================================================================
As technology continues to advance, internet usage are increasing every year, including cellular network internet usage. Cellular operators, as key providers of internet services, must analyze internet consumption patterns within their networks, particularly through traffic data. Such analysis is crucial for understanding user behavior and identifying potential opportunities. However, traffic data consumption is highly dynamic, with patterns changing over time depending on user behavior. This variability poses challenges for operators, requiring advanced methods to analyze traffic data patterns in the time domain. Currently, the operator observed in this study relies on basic descriptive analysis to group potential traffic data consumption by location. This process often involves subjective decision-making, especially in identifying Points of Interest (POI) during national holidays such as Eid al-Fitr, Christmas, and New Year. This approach is time-intensive, particularly when characterizing each location based on its traffic data consumption patterns. Given that traffic data is inherently time-series in nature, there is a need for methods capable of efficiently identifying and analyzing these temporal patterns. This study utilizes time-series clustering methods to analyze traffic data consumption and uncover patterns across 61 districts managed by a cellular operator in Indonesia. Several clustering techniques—K-Means, Partitioning Around Medoids (PAM), and Fuzzy C-Means—were compared, with the Dynamic Time Warping (DTW) algorithm applied to measure pattern similarity. Among the methods evaluated, the combination of DTW and Fuzzy C-Means demonstrated superior performance, achieving the highest silhouette score and the lowest Davies-Bouldin Index.The analysis identified three optimal clusters, each characterized by specific parameters: the weekend-to-weekday traffic ratio, traffic ratios during the Eid al-Fitr, Christmas, and New Year periods, and the year-to-date traffic ratio. The characteristics of each cluster provide valuable insights into the potential for traffic data consumption during specific temporal conditions in the analyzed districts. This approach offers a more robust and efficient method for understanding user behavior and optimizing network resources, particularly during critical time periods.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Cellular Network, Data Traffic, Time-Series Clustering, Traffic Data Pattern, Data trafik, Dynamic Time Warping (DTW), Jaringan Seluler, Pola Data trafik
Subjects:	T Technology > T Technology (General) > T174 Technological forecasting T Technology > T Technology (General) > T57.5 Data Processing T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5101 Telecommunication
Divisions:	Interdisciplinary School of Management and Technology (SIMT) > 61101-Master of Technology Management (MMT)
Depositing User:	Muhammad Fachry Nova
Date Deposited:	02 Feb 2025 06:35
Last Modified:	02 Feb 2025 06:35
URI:	http://repository.its.ac.id/id/eprint/117552

Actions (login required)

View Item