Big Data Analytics and Predictive Intelligence for Optimizing Complex Decision-Making Processes: A Structured Review of Methods, Architectures, and Governance Challenges
Abstract
ABSTRACT
Complex decision-making in modern enterprises, healthcare systems, and supply networks is characterised by high-dimensional data, uncertainty, competing objectives, and compressed response times, conditions under which intuition-led and descriptive-reporting approaches increasingly fail. Big data analytics and predictive intelligence, the combination of scalable data infrastructure, machine learning, and optimisation, promise to convert such data into timely and defensible decisions. This structured narrative review synthesises 25 peer-reviewed studies and landmark publications spanning 1997–2023 to examine how big data analytics supports complex decision-making across five dimensions: conceptual foundations, predictive and prescriptive methods, domain applications, governance and explainability, and deployment engineering. Works were purposively selected using explicit criteria, classified thematically, and appraised for methodological transparency. The synthesis shows that (i) tree-ensemble and deep sequence models dominate predictive tasks, yet accuracy rarely improves decisions unless coupled with prescriptive optimisation; (ii) value realisation depends more on organisational analytics capability than on technology; (iii) explainability and privacy-preserving learning are prerequisites for adoption in regulated domains; and (iv) deployment debt, data drift, and feedback loops threaten long-run reliability. A five-layer integrative framework linking data, processing, prediction, prescription, and governance is proposed, together with research priorities for decision-centric evaluation, causal reasoning, and sustainable analytics operations.
KEYWORDS: Big data analytics, Predictive intelligence, Decision-making, Machine learning, Prescriptive analytics, Explainable AI, Decision support systems, MLOps, Data governance, Structured review
Complex decision-making in modern enterprises, healthcare systems, and supply networks is characterised by high-dimensional data, uncertainty, competing objectives, and compressed response times, conditions under which intuition-led and descriptive-reporting approaches increasingly fail. Big data analytics and predictive intelligence, the combination of scalable data infrastructure, machine learning, and optimisation, promise to convert such data into timely and defensible decisions. This structured narrative review synthesises 25 peer-reviewed studies and landmark publications spanning 1997–2023 to examine how big data analytics supports complex decision-making across five dimensions: conceptual foundations, predictive and prescriptive methods, domain applications, governance and explainability, and deployment engineering. Works were purposively selected using explicit criteria, classified thematically, and appraised for methodological transparency. The synthesis shows that (i) tree-ensemble and deep sequence models dominate predictive tasks, yet accuracy rarely improves decisions unless coupled with prescriptive optimisation; (ii) value realisation depends more on organisational analytics capability than on technology; (iii) explainability and privacy-preserving learning are prerequisites for adoption in regulated domains; and (iv) deployment debt, data drift, and feedback loops threaten long-run reliability. A five-layer integrative framework linking data, processing, prediction, prescription, and governance is proposed, together with research priorities for decision-centric evaluation, causal reasoning, and sustainable analytics operations.
KEYWORDS: Big data analytics, Predictive intelligence, Decision-making, Machine learning, Prescriptive analytics, Explainable AI, Decision support systems, MLOps, Data governance, Structured review