Penerapan Metode SVM-Based Machine Learning Untuk Menganalisa Pengguna Data Trafik Internet (Studi Kasus Jaringan Internet Wlan Mahasiswa Bina Darma)

TL;DR

This study applies SVM-based machine learning using WEKA to classify internet traffic data from campus Wi-Fi hotspots, identifying usage patterns, destination networks, and protocols to improve understanding of network activity.

Contribution

The paper demonstrates the effectiveness of SVM in classifying internet traffic patterns from campus Wi-Fi data, providing insights into network usage and protocol preferences.

Findings

SVM achieved high accuracy in classifying network traffic patterns.

Identified the most frequently accessed destination networks and protocols.

Demonstrated the utility of machine learning for network traffic analysis.

Abstract

Internet usage is an important requirement that supports the performance and activities on campus. To control internet usage, it is necessary to know the distribution of internet usage. By utilizing a number of machine learning algorithms and WEKA software, the research is carried out by observation and taking data from wifi hotspots on campus. The classification method using SVM-Based utilizes the classification method owned by Support Vector Machine (SVM). This study aims to classify data on internet usage so that from this classification can be known destination network, protocol, and bandwidth that are widely accessed at certain times. Internet traffic data is retrieved through Wireshark software. Whereas data processing and data processing of internet traffic is processed by WEKA. The results showed: 1) UBD internet usage in the week I 133,196 users, week II 304,042 users,2) Use of Destination Network 24,150 and Use of Protocol 37,321,3) Destination networks that are often addressed are 172.21.206.143 (the week I) and 172.21.172.234 (week II), protocols that are often used by TCP, and4) SVM method is a good data mining method for classifying network packet patterns so as to produce network traffic classification according to destination network and protocol.