From Theory to Impact: Building Data-Driven Models for Real-World Optimization

This talk highlights recent efforts to build theory- enabled, data-driven models for optimizing complex operational decisions across multiple sectors, with a focus on bridging the gap between rigorous analytical frameworks and real-world implementation. A core emphasis of our work is on developing models that not only leverage data effectively but also provide robust performance in volatile environments where uncertainty and rapid change are inherent. A central methodological advancement lies in the theory of moments, which we extend to develop tractable distributionally robust optimization frameworks. These models allow decision-makers to account for ambiguity in demand, supply, or pricing dynamics by using limited statistical information (e.g., means, variances) without relying on precise distributional assumptions. This is particularly valuable in environments with high volatility, where classical optimization approaches may fail to perform reliably. We apply this advancement in multiple domains, such as airline revenue management, where it informs real-time pricing and offer decisions under uncertain demand, and in logistics and humanitarian operations, where we model resource allocation and routing under unpredictable disruptions. This work reflects the growing importance of operations research (OR) in Singapore’s innovation ecosystem, where data and theory converge to solve practical challenges. It also demonstrates how theory-to-practice translation can create scalable, high-impact decision tools for sectors ranging from aviation to public services, especially in an era defined by volatility and complexity.