Predictive Analytics for Supply Chain Resilience and Logistics Optimisation in Volatile Operating Environments
Abimbola Serifat Oreoluwa *
Department of Cybersecurity, Nexford University, Washington, DC USA.
Betty Ansah
Northeastern University, Portland, ME, USA.
Priscilla Agboada
Vanderbilt University, Accra, Ghana.
*Author to whom correspondence should be addressed.
Abstract
Introduction: Volatile operating environments create demand uncertainty, supplier disruptions, delivery delays, geopolitical shocks, and other events that can negatively affect operational performance and supply chain resilience. Evidence on the contribution of predictive analytics to both logistics optimisation and resilience is still fragmented, and opportunities exist to anticipate these risks and inform timely logistics decision-making. This systematic review aims to collate evidence on the use of predictive analytics to improve supply chain resilience and optimise logistics operations in volatile operating environments, with a specific focus on predictive techniques, applications, outcomes, and implementation challenges.
Methods: A systematic literature review was conducted in accordance with PRISMA 2020. Terms relating to predictive analytics, supply chains and logistics, resilience and risk, and optimisation were used to search Scopus and the Web of Science Core Collection. Eligible studies applied machine learning, deep learning, artificial intelligence (AI)-based predictive modelling, or related forecasting approaches to problems involving uncertainty, disruption, risk, resilience, or optimisation in supply-chain and/or logistics settings. After deduplication, 321 unique records were screened by title and abstract; 30 reports underwent full-text assessment, and 15 studies were included in the synthesis.
Results: The 15 included studies showed substantial methodological and contextual heterogeneity. Applications concentrated on demand forecasting, inventory management, supplier risk, disruption prediction, and product availability. Methods included ensemble learning, deep learning, hybrid models, and adaptive approaches. Predictive performance often improved relative to study-specific comparators, although greater model complexity did not consistently produce better performance. The clearest operational gains were reported when predictions were coupled with optimisation or decision-support processes such as inventory allocation, replenishment, and supplier selection. Evidence relevant to resilience was strongest for anticipation, visibility, response, and recovery, whereas direct evidence for longer-term adaptation remained limited. Common implementation constraints included data quality, data availability, model adaptability, interpretability, and organisational infrastructure.
Conclusion: Predictive analytics can support supply-chain resilience and logistics performance when predictive outputs are translated into timely operational decisions. The evidence remains heterogeneous and context-dependent; future research should prioritise external validation, longitudinal assessment, and direct measurement of resilience outcomes.
Keywords: Predictive analytics, machine learning, supply-chain resilience, logistics optimisation, disruption prediction, demand forecasting