Kombinasi Algoritma Support Vector Machine Dan Lexicon Based Pada Analisis Sentimen Twitter (Studi Data: PP Tapera No 25 Tahun 2020)

Rindu, Hafil Muhammadi (2021) Kombinasi Algoritma Support Vector Machine Dan Lexicon Based Pada Analisis Sentimen Twitter (Studi Data: PP Tapera No 25 Tahun 2020). Undergraduate Thesis thesis, Institut Teknologi Telkom Purwokerto.

	Text Cover.pdf Download (457kB)
	Text Abstract .pdf Download (8kB)
	Text Abstrak .pdf Download (8kB)
	Text BAB I.pdf Download (91kB)
	Text BAB II.pdf Download (284kB)
	Text BAB III.pdf Download (459kB)
	Text BAB IV.pdf Restricted to Registered users only Download (319kB) | Request a copy
	Text BAB V.pdf Download (10kB)
	Text Daftar Pustaka.pdf Download (156kB)
	Text Lampiran.pdf Restricted to Registered users only Download (173kB) | Request a copy

Abstract

Data from the Ministry of Public Works and Public Housing (Kementrian PUPR) in 2019, around 81 million millennials do not own a home. Government Regulation Number 25 of 2020 concerning the Implementation of Public Housing Savings or commonly called PP 25 Tapera of 2020 is the government's effort so that Indonesian people can own houses. Tapera is a term deposit by workers for housing finance, or can be returned again after the term expires. When this regulation was enacted, there were many public responses regarding this regulation. Based on the community's response, research can be applied to see public sentiment regarding this regulation. The method used for this research is support vector machine (SVM) because SVM has a good level of accuracy. Implementation of SVM in addition to requiring training data, also requires labeling, and labeling manually takes a lot of time. The labeling method used is lexicon-based. This combination using lexicon-based to determine the sentiment value or data label and the lexicon result data is used as labeling data for the Support Vector Machine classifier input. The combination makes lexicon-based a means to transfer learning to SVM by facilitating the labeling process. The research process begins with collecting research data from Twitter social media, then the preprocessing process to convert raw and unstructured data into data that is ready to use, then labeling the data with lexicon-based, then weighting with TF-IDF, processing using SVM, and the evaluation process algorithm performance model with confusion matrix. The results showed that the combination of lexicon-based and SVM worked well. Lexicon-based managed to label 640 tweet data and the labeling data was used by SVM. SVM managed to get an accuracy value of 83.08% for linear kernel. The linear kernel also has the best results for the recall value, and f1-score with values of 63.19%, 64.56%, respectively. As for the precision value, the rbf kernel has the best result with 75.29%. Kata Kunci: Analisa Sentimen, Tapera, Support Vector Machine, Lexicon Based

Actions (login required)

View Item