Opinion Mining Pengguna X (Twitter) terhadap Putusan MK Tentang Pemilu 2024 dengan model Naive Bayes Classifier dan Gradient Boosted Machines

Permana, Indra Surya
Juliane, Christina

TECHNOVATAR
Vol 2 No 4 (2024): OKTOBER

This study explores sentiment analysis of Twitter reactions to the Constitutional Court's verdict on Indonesia's 2024 election, using Naive Bayes and Gradient Boosted Machines (GBM). Aimed at assessing the efficacy of these models in categorizing sentiments as positive, negative, or neutral, the data were processed using TF-IDF vectorization. Results reveal that Naive Bayes excels in identifying positive sentiments with high precision and recall, yet struggles with negative and neutral sentiments. Conversely, GBM demonstrates a more balanced performance across all categories, particularly in detecting neutral sentiments. This highlights the strengths and weaknesses of each model, suggesting Naive Bayes for scenarios where positive sentiment detection is critical, and GBM for situations requiring nuanced sensitivity across diverse sentiments. This research offers insights into the strategic selection of models for sentiment analysis based on specific needs.