Perbandingan Metode Naive Bayes dan Support Vector Machine (SVM) untuk Opini Publik Mengenai Gaji DPR RI di Media Sosial

TABALA, Elga Afliana (2026) Perbandingan Metode Naive Bayes dan Support Vector Machine (SVM) untuk Opini Publik Mengenai Gaji DPR RI di Media Sosial. Undergraduate thesis, Universitas Katolik Widya Mandira.

	Text ABSTRAK.pdf Download (1MB)
	Text BAB I.pdf Download (343kB)
	Text BAB II.pdf Restricted to Repository staff only Download (429kB)
	Text BAB III.pdf Restricted to Repository staff only Download (575kB)
	Text BAB IV.pdf Restricted to Repository staff only Download (1MB)
	Text BAB V.pdf Download (132kB)
	Text DAFTAR PUSTAKA & SURAT BEBAS PLAGIASI.pdf Download (224kB)

Abstract

Analyzes public sentiment regarding Indonesian House of Representatives (DPR RI) salaries/allowances on X (Twitter) and compares the performance of Naïve Bayes and Support Vector Machine (SVM) for classifying sentiments into positive, negative, and neutral. The data were collected through a crawling process using tweet-harvest@latest during August 25, 2025–October 29, 2025, resulting in 1,500 Indonesian-language tweets. Sentiment labels were assigned manually into three classes (positive=1, negative=2, neutral=3), after which the data underwent cleaning and preprocessing (case folding, punctuation removal, stopword removal, tokenizing, and normalization/stemming) and were represented using TF– IDF weighting. The dataset was split into 80% training data (1,032 tweets) and 20% testing data (259 tweets) for model training and evaluation. The results show that public sentiment is predominantly negative at 89.47% (1,500 items), followed by positive at 6.20% (80 items) and neutral at 4.34% (56 items). Performance evaluation indicates that Naïve Bayes achieved 76% accuracy, while SVM achieved 91% accuracy, demonstrating that SVM provides better overall performance on the study dataset

Actions (login required)

View Item