Statistical and Machine Learning Approaches to Production Optimization in the Brewery Industry

ABSTRACT: Production collapse in brewery operations is a major industrial challenge marked by sustained declines in output, efficiency, and capacity utilization due to interacting technical, operational, managerial, and external constraints. This systematic review synthesizes existing literature on the root causes of production decline in the brewery and beverage industry, with emphasis on developing economies. Guided by the PRISMA framework and drawing from major scientific databases, the study examines empirical evidence on critical production bottlenecks. The review compares traditional mathematical models with advanced Machine Learning (ML) techniques for root cause identification, highlighting their complementary strengths in interpretability and predictive accuracy. It further evaluates optimization and what-if scenario analysis as decision-support tools for translating predictive insights into practical production improvements. Evidence shows that scenario-based optimization can enhance output, reduce downtime, and improve resource allocation in brewery systems. Despite progress, gaps remain, particularly the absence of integrated root-cause, ML, and optimization frameworks and limited validation rigor. By consolidating fragmented findings and outlining future research directions, this review provides a structured foundation for developing robust, data-driven productivity recovery strategies and strengthening sustainable performance in brewery operations.