Analysis of Indonesian News Patterns and Clustering Using the TF-IDF and K-Means Methods with PCA Visualization

Abstract

The rapid growth of online news in Indonesia has been driven by the expansion of digital media and social networking platforms. The vast amount of news data requires automated analytical methods to efficiently group and identify topic patterns. This study presents an analysis and clustering of Indonesian news articles using a hybrid approach combining Term Frequency–Inverse Document Frequency (TF-IDF) and K-Means Clustering, with Principal Component Analysis (PCA) applied for visualization. The research utilizes the Indonesian News Dataset (Kaggle), consisting of thousands of articles from diverse categories such as politics, economics, crime, sports, entertainment, government, and general topics. The methodology includes text preprocessing, TF-IDF vectorization, K-Means clustering, and PCA-based dimensionality reduction to visualize the cluster distribution. Experimental results demonstrate that the proposed combination of TF-IDF and K-Means effectively groups news articles with high consistency between dominant keywords and actual category labels. The PCA visualization reveals clear separations between clusters, particularly in the politics, economics, and sports categories. This research contributes to the implementation of unsupervised learning techniques for Indonesian-language big data text analysis