Course Objective & Scope
Machine learning predicts what will happen, but choosing the best action under budget, capacity, shift, fleet or SLA constraints requires optimisation. This programme covers defining decision variables, objectives and constraints, linear and mixed-integer programming (LP/MILP), using Google OR-Tools and Pyomo/HiGHS, and predict-then-optimize architectures where ML predictions feed the decision model.
Fragility caused by point predictions in optimisation is addressed through scenario-based stochastic optimisation and contextual bandits. Participants build a working decision prototype and sensitivity report on an allocation, scheduling or pricing problem from their organisation.
Who Should Attend?
- Data scientists and operations research (OR) specialists
- Supply chain, logistics, manufacturing and revenue management analysts
- Engineers building decision-support and pricing engines
- Prerequisite: Python and basic linear algebra/probability
Key Takeaways
- Formulate a business problem with decision variables, constraints and an objective
- Solve LP, MILP and constraint programming models with OR-Tools and Pyomo
- Build end-to-end pipelines that turn ML predictions into optimisation inputs
- Reduce sensitivity to forecast error using scenario and robust optimisation
- Quantify optimisation gains against heuristic rules through simulation
Curriculum
From Prediction to Decision: Mathematical Formulation
- Decision variables, objectives, hard and soft constraints
- LP, MILP and linearising common non-linear structures
Modelling with OR-Tools and Pyomo
- Assignment, capacity allocation, shift scheduling and VRP
- HiGHS/CBC solvers, optimality gap and time limits
Predict-Then-Optimize Architecture
- Feeding demand, churn or elasticity predictions into the solver
- Forecast error vs decision regret
Decision-Making Under Uncertainty
- Scenario-based stochastic programming and chance constraints
- Contextual bandits and the explore-exploit trade-off
Simulation, Sensitivity and Production Readiness
- Backtesting against current business rules and shadow prices
- Constraint-violation monitoring and explainable decision outputs
Other Courses in This Category
-
Mastering tabular gradient boosting with XGBoost, LightGBM and CatBoost: tree-growth strategies, native categorical and missing-value handling, ranking/quantile/custom objectives, monotonic constraints, Optuna and fast inference.
- Live Course
- 5 modules
-
Automated model selection, hyperparameter search and multi-layer stacking with AutoGluon, FLAML and Optuna, covering compute budget management, leakage checks and moving from rapid baselines to production.
- Live Course
- 5 modules
-
Glass-box models (EBM, GAM), TreeSHAP/KernelSHAP, ALE, counterfactual explanations (DiCE), uncertainty quantification and regulatory compliance for producing auditable, actionable explanations in high-stakes decisions.
- Live Course
- 5 modules
-
Moving from correlation to causation with DAGs, propensity scores, Double Machine Learning (EconML/DoWhy), heterogeneous treatment effects (CATE) and Difference-in-Differences to measure price, campaign and policy impact.
- Live Course
- 5 modules
-
Collaborative model training across institutions or edge devices without centralising raw data: Flower framework, FedAvg/FedProx, non-IID data, differential privacy (Opacus) and secure aggregation.
- Live Course
- 5 modules
-
Intro to Machine Learning - Regression Models
Hands-on workshop covering Linear, Ridge, Lasso, and Polynomial regression models, feature engineering, MSE/RMSE/MAE/R² evaluation metrics, and regularization techniques.
- Live Course
- 3 modules
Get in touch
