PT PLN (Persero) sebagai Badan Usaha Milik Negara (BUMN) memiliki peran strategis dalam penyediaan listrik nasional. Dinamika lingkungan eksternal, seperti kebijakan pemerintah, perkembangan teknologi, hingga isu lingkungan, dapat memengaruhi operasional dan strategi bisnis perusahaan. Penelitian ini bertujuan mengimplementasikan algoritma Machine Learning khususnya Support Vector Machine (SVM) untuk mengklasifikasikan berita daring terkait PT PLN berdasarkan enam dimensi PESTEL (Political, Economic, Social, Technological, Environmental, Legal). Dataset berjumlah 727 berita diambil dari Antara.com dan Kompas.com periode 2023–2025 melalui teknik web scraping. Metodologi penelitian menggunakan CRISP-DM dengan tahapan business understanding, data understanding, data preparation, modeling, evaluation, dan deployment. Proses preprocessing meliputi case folding, tokenization, stopword removal, stemming, serta pembobotan kata dengan TF-IDF. Model SVM dievaluasi menggunakan metrik akurasi, presisi, recall, dan F1-score. Hasil menunjukkan akurasi sebesar 69,89% dengan kategori Technological menjadi faktor eksternal paling dominan dalam pemberitaan terkait PLN. Penelitian ini diharapkan membantu PT PLN dalam memantau isu eksternal secara otomatis, meningkatkan efisiensi analisis informasi, dan mendukung pengambilan keputusan strategis berbasis data.

PT PLN (Persero) is a state-owned enterprise (SOE) with a strategic role in national electricity provision. External environmental dynamics, such as government policies, technological advancements, and environmental issues, may influence the company’s operations and business strategies. This study aims to implement the Support Vector Machine (SVM) algorithm to classify online news related to PT PLN into six PESTEL dimensions (Political, Economic, Social, Technological, Environmental, Legal). The dataset consists of 727 news articles collected from Antara.com and Kompas.com (2023 2025) using web scraping. The research applies the CRISP-DM methodology, comprising business understanding, data understanding, data preparation, modeling, evaluation, and deployment. Text preprocessing includes case folding, tokenization, stopword removal, stemming, and TF-IDF weighting. The SVM model was evaluated using accuracy, precision, recall, and F1-score metrics. Results show an accuracy of 69.89%, with the Technological category being the most dominant external factor in PLN-related news. This study is expected to assist PLN in automatically monitoring external issues, improving information analysis efficiency, and supporting data-driven strategic decision making.