Perbandingan Metode Naïve Bayes dan K-Nearest Neighbor dalam Analisis Sentimen Terhadap Program Makan Bergizi Gratis

WATTIMENA, Christy Inda (2026) Perbandingan Metode Naïve Bayes dan K-Nearest Neighbor dalam Analisis Sentimen Terhadap Program Makan Bergizi Gratis. Undergraduate thesis, Universitas Katolik Widya Mandira.

	Text ABSTRAK.pdf Download (645kB)
	Text BAB I.pdf Download (271kB)
	Text BAB II.pdf Restricted to Repository staff only Download (425kB)
	Text BAB III.pdf Restricted to Repository staff only Download (517kB)
	Text BAB IV.pdf Restricted to Repository staff only Download (1MB)
	Text BAB V.pdf Download (245kB)
	Text DAFTAR PUSTAKA DAN SURAT BEBAS PLAGIAT.pdf Download (350kB)

Abstract

The Free Nutritious Meal (MBG) Program is a government social policy that has elicited various public responses on social media. This study aims to analyze public sentiment toward the MBG Program and compare the performance of Naïve Bayes (NB) and K-Nearest Neighbor (KNN) algorithms in text-based sentiment classification. Research data was collected from the X platform using keywords related to MBG during the period of August 1, 2025, to September 30, 2025, with a total of 15,854 Indonesian-language tweets. After cleaning and preprocessing, 14,169 valid tweets were used for modeling. Preprocessing stages included case folding, punctuation removal, stopword removal (with normalization), tokenizing, and stemming. Furthermore, text features were formed using TF-IDF weighting, and the data was split into training and testing sets with a 70:30 ratio using the Stratify technique. Classification was conducted into three sentiment classes: negative, neutral, and positive. The results showed that the sentiment distribution was dominated by negative sentiment at 55.09% (7,805 data), followed by neutral sentiment at 23.25% (3,294 data) and positive sentiment at 21.67% (3,070 data). Performance evaluation using accuracy, precision, recall, and F1-score metrics showed that Naïve Bayes achieved an accuracy of 72%, while KNN achieved an accuracy of 71%. Considering the overall accuracy and performance characteristics per class, Naïve Bayes provided better performance on this research dataset.

Actions (login required)

View Item