📅 23 February 2024
DOI: 10.35671/telematika.v17i1.2824
Comparative Analysis of Classification Methods in Sentiment Analysis: The Impact of Feature Selection and Ensemble Techniques Optimization
Telematika
Universitas Amikom Purwokerto
📄 Abstract
Optimizing classification methods (forward selection, backward elimination, and optimized selection) and ensemble techniques (AdaBoost and Bagging) are essential for accurate sentiment analysis, particularly in political contexts on social media. This research compares advanced classification models with standard ones (Decision Tree, Random Tree, Naive Bayes, Random Forest, K-NN, Neural Network, and Generalized Linear Model), analyzing 1,200 tweets from December 10-11, 2023, focusing on "Indonesia" and "capres." It encompasses 490 positive, 355 negative, and 353 neutral sentiments, reflecting diverse opinions on presidential candidates and political issues. The enhanced model achieves 96.37% accuracy, with the backward selection model reaching 100% accuracy for negative sentiments. The study suggests further exploration of hybrid feature selection and improved classifiers for high-stakes sentiment analysis. With forward feature selection and ensemble method, Naive Bayes stands out for classifying negative sentiments while maintaining high overall accuracy (96.37%).
🔖 Keywords
#Sentiment analysis; Classification methods; Feature selection; Ensemble techniques; Text mining
ℹ️ Informasi Publikasi
Tanggal Publikasi
23 February 2024
Volume / Nomor / Tahun
Volume 17, Nomor 1, Tahun 2024
📝 HOW TO CITE
Defit, Sarjon; Universitas Putra Indonesia “YPTK” Padang; Windarto, Agus Perdana; STIKOM Tunas Bangsa; Alkhairi, Putrama; STIKOM Tunas Bangsa; , "Comparative Analysis of Classification Methods in Sentiment Analysis: The Impact of Feature Selection and Ensemble Techniques Optimization," Telematika, vol. 17, no. 1, Feb. 2024.
ACM
ACS
APA
ABNT
Chicago
Harvard
IEEE
MLA
Turabian
Vancouver
Download Citation
0
Sitasi
Cek Google Scholar