Literatur Review: Penggunaan Algoritma Machine Learning untuk Optimalisasi Rantai Pasokan

Abstract

Dalam era bisnis yang semakin kompleks, optimalisasi rantai pasokan menjadi elemen kunci untuk meningkatkan efisiensi operasional, mengurangi biaya, dan menghadapi ketidakpastian pasar. Penelitian ini bertujuan mengkaji kontribusi teknologi machine learning (ML) dalam optimalisasi rantai pasok melalui pendekatan literature review terhadap 15 jurnal bereputasi internasional dalam sepuluh tahun terakhir (2015–2025). Hasil kajian menunjukkan bahwa ML berperan signifikan dalam tiga aspek utama: efisiensi operasional, akurasi prediksi permintaan, dan pengelolaan inventaris. Berbagai algoritma seperti Neural Networks, LSTM, XGBoost, SVM, Deep Q-Network, hingga metode meta-learning dan reinforcement learning digunakan untuk meningkatkan kecepatan respons, ketepatan prediksi tren pasar, serta pengelolaan stok berbasis data real-time. Selain itu, integrasi ML dengan teknologi seperti IoT dan RFID juga memperkuat efisiensi sistem distribusi dan deteksi anomali. Meskipun tantangan seperti keterbatasan data, infrastruktur teknologi, dan keamanan informasi masih menjadi hambatan, penelitian ini memperlihatkan bahwa penerapan ML menawarkan solusi strategis untuk membangun rantai pasok yang adaptif, cerdas, dan berkelanjutan di era digital. Temuan ini memberikan arah bagi penelitian dan implementasi lebih lanjut dalam pengembangan supply chain berbasis teknologi cerdas.

In an increasingly complex business era, supply chain optimization became a key element to improve operational efficiency, reduce costs, and address market uncertainties. This study aimed to examine the contribution of machine learning (ML) technology in optimizing the supply chain through a literature review approach on 15 internationally reputable journals over the past ten years (2015–2025). The review results showed that ML played a significant role in three main aspects: operational efficiency, demand forecasting accuracy, and inventory management. Various algorithms such as Neural Networks, LSTM, XGBoost, SVM, Deep Q-Network, as well as meta-learning and reinforcement learning methods, were used to enhance response speed, market trend prediction accuracy, and real-time data-based stock management. In addition, the integration of ML with technologies such as IoT and RFID also strengthened distribution system efficiency and anomaly detection. Although challenges such as data limitations, technological infrastructure, and information security remained obstacles, this study demonstrated that the application of ML offered strategic solutions to build an adaptive, intelligent, and sustainable supply chain in the digital era. These findings provided direction for further research and implementation in the development of technology-based intelligent supply chains.