Overview
Data-Driven Optimization focuses on learn decisions from data, not just parameters. In the map of OR, it connects Contextual optimization, DRO, Decision-focused learning to decisions that must be modeled, solved, explained, and revised as evidence changes.
Contextual optimization, predictive stochastic programming, decision-focused learning, and distributionally robust optimization connect ML and OR. The practical use case is clearest in Personalized medicine, Routing, Pricing, Inventory, where the method helps turn constraints and tradeoffs into a decision artifact someone can inspect.
Core ideas
Contextual optimization
Contextual optimization is a core checkpoint for Data-Driven Optimization: define it concretely, attach units or rules where possible, and test whether stakeholders interpret it the same way.
DRO
DRO is a core checkpoint for Data-Driven Optimization: define it concretely, attach units or rules where possible, and test whether stakeholders interpret it the same way.
Decision-focused learning
Decision-focused learning is a core checkpoint for Data-Driven Optimization: define it concretely, attach units or rules where possible, and test whether stakeholders interpret it the same way.
Policy learning
Policy learning is a core checkpoint for Data-Driven Optimization: define it concretely, attach units or rules where possible, and test whether stakeholders interpret it the same way.
How to use it
- 1Start with Personalized medicine: write the decision, time horizon, actors, and objective in operational language.
- 2Translate the problem into Contextual optimization, DRO, and Decision-focused learning; define units and data sources for each one.
- 3Build a small instance of Data-Driven Optimization that can be solved or simulated by hand inspection before using full production data.
- 4Compare the recommendation against a baseline policy, not just against mathematical optimality.
- 5Document assumptions, sensitivity results, and the conditions under which the recommendation should be revisited.
Applications
- Personalized medicine: compare feasible policies, quantify the operating tradeoffs, and make the assumptions behind the recommendation visible.
- Routing: compare feasible policies, quantify the operating tradeoffs, and make the assumptions behind the recommendation visible.
- Pricing: compare feasible policies, quantify the operating tradeoffs, and make the assumptions behind the recommendation visible.
- Inventory: compare feasible policies, quantify the operating tradeoffs, and make the assumptions behind the recommendation visible.
Common pitfalls
- Applying Data-Driven Optimization because the label sounds appropriate while leaving the actual decision boundary vague.
- Treating Contextual optimization as a technical detail instead of a modeling choice that affects the recommendation.
- Reporting one answer without showing sensitivity to demand, capacity, costs, or behavioral assumptions.
- Ignoring implementation details such as data quality, explainability, ownership, and how users will override bad recommendations.
Resources
- Contextual Optimization Survey
Topic-specific source curated for Data-Driven Optimization.
- Algorithms for Decision Making
Open book covering planning, MDPs, reinforcement learning, and decision algorithms.
- CVXPY Short Course
Hands-on examples for disciplined convex modeling in Python.
- INFORMS Ethics Guidelines
Professional ethics guidance for analytics, models, and decision systems.