45-950 AI for Smarter Operational Decisions
6 units, Carnegie Mellon University
This course explores how Artificial Intelligence (AI) methods can enhance operational decision- making. It builds on Operations Research (OR) — the discipline of analytics-based decision- making — and integrates Machine Learning (ML) with optimization to improve business decisions. Traditional OR methods have separated the process of going from data to decisions into sequentially solved stages. For example, we first predict the demand, after which we will optimize a production schedule against the prediction. This separation potentially weakens the outcomes but has been necessary to keep each step manageable in terms of model size and computation. Bridging the stages by integrating their respective model can lead to better outcomes, e.g., cost savings, higher productivity, or lower risk. The recent development and application of machine learning methods to business problems have spawned a closer integration of learning (ML) and decision making (OR). These range from incorporating optimization methods into ML (e.g., via decision loss) to embedding linearized predictive ML model into optimization. These topics are still under active development by the research community and industry. We will explore some of the more established integrations of ML and OR; where business cases are available, we will use them. Otherwise, we will assess the potential business value on synthetic datasets. The lectures combine theory and methodology with hands-on cases. We will use Python to build our ML models and our Gurobi optimization. The course will include a multi-week project to be done in small groups as well as four individual homework assignments.
Prerequisites: none listed
Offered: Spring 2027, Spring 2026
Courseweave needs JavaScript to show sections, workloads and planning tools.