KLUSTERING TOPIK PADA KOLOM KOMENTAR INSTAGRAM TENTANG KABINET MERAH PUTIH MENGGUNAKAN METODE K-MEANS
Clustering Topics in Instagram About Cabinet Merah Putih Using the K-Means Method
Abstract
This research attempts to determine the primary themes that Indonesians talked on President Prabowo Subianto's "Merah Putih" cabinet by using clustering analysis of Instagram comments. Using crawling data from the Instagram platform with the hashtag #KabinetMerahPutih, comments were gathered using the K-Means Clustering approach. Prior to the data being analyzed to create five clusters, the cleaning and pre-processing procedure, which included tokenization with IndoBERT and dimensionality reduction using Principal Component Analysis (PCA), was able to greatly improve the clustering quality, with the Silhouette Coefficient value rising from 0.010 to 0.200. Out of 23.780 initial data, 9.320 clean data were processed for this investigation. The findings demonstrate that the K-Means algorithm can group comments according to pertinent themes and offer profound understanding of support more responsive public policy analysis.