Choosing the best way to make decisions with uncertain contexts
Contextual Decision Making: Predictive or Generative
Machine Learning
Summary
Making decisions based on uncertain information can be done by predicting outcomes first or by generating possible scenarios and then choosing the best option. The paper compares these two approaches, showing that each works better depending on the smoothness of the problem and how much data is available. Sometimes predicting is simpler but less flexible, while generating scenarios keeps problems easier to solve but needs more data and computation. The authors provide guidance on when to use each method but find no single best choice for all situations.
What this means in practice
- •For supply chain managers: Improve inventory decisions under demand uncertainty by choosing whether to predict demand directly or generate demand scenarios for optimization.
- •For financial risk teams: Design better portfolio strategies by understanding when to rely on predicted returns or on simulated market scenarios for risk assessment.
Authors
L. Jeff Hong, Weihuan Huang, Xinyao Li
Abstract
Contextual decision making uses observed covariates to guide decisions under uncertainty. Two natural approaches are predict-then-optimize (PtO), which learns the conditional expected objective, and generate-then-optimize (GtO), which learns conditional uncertainty and optimizes using generated scenarios. We study when each approach is preferable by examining how the choice of learning target affects statistical accuracy and optimization tractability. Although PtO learns a narrower target, our regret bounds show how differences in smoothness can offset GtO's additional dimensional burden. Exploiting smoothness also raises an optimization tradeoff: within the class of nonnegative-weight PtO methods studied, a lower bound limits the gains from additional smoothness, while more flexible methods may introduce signed weights and lose the guarantee of preserving convexity. GtO preserves scenario-wise convexity, but requires learning a richer object and generating scenarios for optimization. Synthetic experiments and a real case study illustrate how these tradeoffs vary with data availability and problem structure. Together with the scope for reusing a model across tasks, our findings provide practical guidance for choosing between the approaches without identifying a universal winner.