Integration of Machine Learning and OR combines predictive analytics with prescriptive optimization. ML provides: demand forecasts, parameter estimates, pattern recognition feeding OR models. OR provides: optimal decisions under constraints, interpretable solutions, fairness guarantees. Approaches: (1) Predict-then-Optimize - train ML model, use predictions in optimization; (2) Prescriptive Analytics - optimize decisions accounting for prediction uncertainty; (3) Learning-to-Optimize - ML learns to solve optimization problems faster. Applications: inventory management with ML demand forecasts, dynamic pricing, workforce scheduling, and energy systems. Emerging area: differentiable optimization layers in neural networks, enabling end-to-end learning. Challenges: balancing prediction accuracy with decision quality, handling distribution shift.