Applications of Machine Learning in Solving Optimisation Problems: Trends, Methods, and Practical Use Cases

Abstract

Optimisation problems are central to diverse domains such as engineering, logistics, finance, and healthcare, yet traditional methods often rely on handcrafted heuristics and rigid mathematical programming, limiting their scalability and adaptability. Recent advances in machine learning (ML) have introduced transformative approaches that can address these limitations by offering adaptive, data-driven strategies. The purpose of this study is to provide a comprehensive analysis of ML-based optimisation methods, focusing on emerging trends, methodological distribution, performance comparisons, application domains, and the impact of research. The methodology employed integrates a systematic review of recent literature with comparative evaluations illustrated through six figures, covering publication trends (2019–2025), performance metrics, method usage distribution, domain-specific applications, workflow performance, and a research impact matrix. Results show a significant rise in the use of reinforcement learning, deep learning, and hybrid methods, with performance improvements of up to 85% in accuracy, a 3.2-fold speed enhancement, and a 67% cost reduction compared to traditional approaches. Domain analysis reveals that engineering and logistics are leading areas of adoption, while healthcare, finance, and aerospace represent emerging yet impactful applications. The impact matrix further highlights that no single method dominates all domains, reinforcing the importance of hybrid and adaptive strategies. The novelty of this study lies in its integrative framework, which combines trend analysis, performance evaluation, and impact mapping, offering a holistic understanding of how ML is reshaping optimisation research and practice.