Web users segmentation using Genetic K-Means Algorithm

Roiha, Nur Ulfatur (2017) Web users segmentation using Genetic K-Means Algorithm. Masters thesis, Institut Teknologi Sepuluh Nopember.

[img]
Preview
Text
2214206701-Master_Theses.pdf - Published Version

Download (1MB) | Preview

Abstract

Kebutuhan dan ketergantungan terhadap internet semakin hari semakin meningkat yang menyebabkan trafik internetpun meningkat. Dengan trafik yang semakin tinggi, maka akses/koneksi internet akan semakin berat/lambat. Sehingga perlu diketahui bagaimana pola trafik internet yang ada selama ini. Pola tersebut berguna untuk dijadikan dasar kebijakan manajemen koneksi internet untuk saat sekarang dan diwaktu yang akan datang. Penelitian ini betujuan untuk melakukan segmentasi pengguna web berdasarkan pola perilaku kunjungan menggunakan metode Genetic K-Means Algorithm. Hasil cluster divalidasi menggunakan metode Silhouette Index. Dengan menggunakan Silhoulette Index dapat diketahui bahwa cluster yang dihasilkan oleh Genetic K-Means Algorithm mengalami peningkatan kualitas sebesar 28,71% lebih baik dibandingkan dengan cluster yang dihasilkan oleh K-Means. Hal ini berarti Genetic K-Means Algorithm bisa mendapatkan cluster yang lebih homogen dan memiliki heterogenitas yang tinggi antar clusternya dibandingkan dengan K-Means ================================================================================================================== Dependance and demand for internet is increasing that causes the increase of internet traffic in daily needs base. The higher traffic affects to internet access connection that will become heavier or slower. Thus, it is necessary to know how the internet traffic pattern occurs every day. The pattern brings an advantage in order to make internet connection management policy for the present and the future time conditions. The research aims is to create web user segmentation based on a web behavior pattern by using Genetic K-Means Algorithm. The cluster result is validated by using Silhouette Index Method that showed that the clusters generated by Genetic K-Means Algorithm has increased for its quality by 28.71% is better than the clusters generated by K-Means. This also has a meaning that Genetic K-Means Algorithm can obtain more homogeneous clusters and have high heterogeneity among inter-clusters compared with K-Means.

Item Type: Thesis (Masters)
Uncontrolled Keywords: Genetic K-Means Algorithm; Silhouette index; Segmentasi
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering
T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105.546 Computer algorithms
T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5101 Telecommunication
Divisions: Faculty of Industrial Technology > Electrical Engineering > (S2) Master Theses
Depositing User: NUR ULFATUR ROIHA
Date Deposited: 24 Mar 2017 04:21
Last Modified: 24 Mar 2017 04:21
URI: http://repository.its.ac.id/id/eprint/2614

Actions (login required)

View Item View Item