Perbandingan Metode Klasifikasi Support Vector Machine (SVM) dan Extreme Gradient Boosting (XGBOOST) Pada Klasifikasi Sentimen Aplikasi Paylater

Zahrah, Salma Fitria Fatimatuz (2023) Perbandingan Metode Klasifikasi Support Vector Machine (SVM) dan Extreme Gradient Boosting (XGBOOST) Pada Klasifikasi Sentimen Aplikasi Paylater. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 06311940000070-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (3MB) | Request a copy

Abstract

Perkembangan financial technology (fintech) di Indonesia sangat pesat. Salah satu dari perkembangan fintech adalah sistem Buy Now-Pay Later atau yang biasa disebut paylater merupakan pembayaran yang ditunda, dengan kata lain seseorang dapat membeli barang saat ini tanpa membayar langsung namun sebagai gantinya mereka membayar tiap bulan beserta bunganya. Sistem paylater sama seperti sistem pada kartu kredit. Contoh aplikasi yang memberikan layanan paylater adalah Kredivo. Suatu layanan akan menghasilkan respon dari pengguna yaitu berupa ulasan. Berdasarkan ulasan tersebut dapat di klasifikasikan berdasarkan sentimen positif dan negatif. Penelitian ini menggunakan dua metode klasifikasi untuk membandingkan ketepatan klasifikasi antara metode Support Vector Machine dan Extreme Gradient Boosting. Penelitian ini dilakukan analisis klasifikasi menggunakan metode Support Vector Machine dengan dua jenis kernel yaitu Linear dan Radial Basis Function (RBF). Pada analisis Support Vector Machine dengan kernel Linear membutuhkan parameter cost (C), dimana nilai parameter C yang akan diuji coba adalah 0,5; 0,75; 1; 10; dan 100. Pada analisis Support Vector Machine dengan kernel RBF menggunakan parameter C dan γ, dimana nilai parameter γ yang akan diuji coba adalah 0,005; 0,05; 0,1; 0,5 dan 0,75. Pada analisis klasifikasi metode Extreme Gradient Boosting dilakukan dengan hyperparameter tuning dengan bantuan metode gridsearch. Hasil dari penelitian ini menunjukkan bahwa metode Support Vector Machine Non-linear dengan kernel RBF parameter C=1 dan γ = 0,75 memiliki ketepatan klasifikasi yang lebih baik daripada Support Vector Machine Linear dan Extreme Gradient Boosting. Dengan hasil rata-rata akurasi, F-score, dan AUC sebesar 94,45%; 96,18% dan 92,39%.
================================================================================================================================
The development of financial technology (fintech) in Indonesia is very rapid. One of the developments in fintech is the Buy Now-Pay Later system or what is commonly called paylater, which is a delayed payment, in other words, someone can buy goods at this time without paying directly but instead they pay each month along with interest. The paylater system is the same as the credit card system. An example of an application that provides paylater services is Kredivo. A service will generate a response from the user in the form of a review. Based on these reviews it can be classified based on positive and negative sentiments. This study uses two classification methods to compare the classification accuracy between the Support Vector Machine and Extreme Gradient Boosting methods. In this study, classification analysis was carried out using the Support Vector Machine method with two types of kernels, namely Linear and Radial Basis Function (RBF). In the Support Vector Machine analysis with the Linear kernel, the cost parameter (C) is required, where the C parameter value to be tested is 0.5; 0.75; 1; 10; and 100. In the Support Vector Machine analysis with the RBF kernel using parameters C and γ, where the value of parameter γ to be tested is 0.005; 0.05; 0.1; 0.5 and 0.75. In the classification analysis of the Extreme Gradient Boosting method, hyperparameter tuning is carried out with the help of the gridsearch method. The results of this study indicate that the Non-linear Support Vector Machine method with RBF kernel parameters C=1 and γ = 0.75 has better classification accuracy than the Linear Support Vector Machine and Extreme Gradient Boosting. With an average accuracy, F-score, and AUC of 94.45%; 96.18% and 92.3%.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	classification, sentiment, SVM, review, XGBOOST, klasifikasi, sentimen, ulasan
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Q Science > QA Mathematics > QA353.K47 Kernel functions (analysis)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Salma Fitria Fatimatuz Zahrah
Date Deposited:	20 Jul 2023 05:28
Last Modified:	20 Jul 2023 05:28
URI:	http://repository.its.ac.id/id/eprint/98695

Actions (login required)

View Item