A Gentle Introduction to Stochastic Programming

Stochastic programming provides a framework for making optimal decisions when input data is uncertain, contrasting with standard linear programming that assumes fixed values. It addresses the challenge of ill-defined problems where constraints depend on random variables. Four primary methods are detailed: robust optimization for worst-case scenarios, chance constraints for probabilistic feasibility, two-stage recourse models for deciding then correcting, and multi-stage recourse models for sequential decision-making. These concepts are explained using the example of a fashion company determining production levels based on uncertain future demand.