Week 3-4 - vulnerability analysis

Week 3-4 - Deep Dive into Scenario Discovery, Sensitivity Analysis, and Uncertainty Management

Welcome to Weeks 3-4 of the course! In the upcoming weeks, you will be exposed to various techniques that will help you analyze and manage uncertainties in your models more effectively. You will delve deeper into scenario discovery, sensitivity analysis, and explore different methods for understanding model behavior in the presence of uncertain parameters. The assignments will build on the topics from previous weeks, applying the techniques learned to the lake problem and predator-prey models.

Throughout Weeks 3-4, you will:

1. Learn about scenario discovery and replicate results from influential papers in decision-making under uncertainty. You will apply the Patient Rule Induction Method (PRIM) to identify policy vulnerabilities, develop Shared Socio-economic Pathways (SSPs), and design adaptive decision-making strategies.
2. Continue working with the lake problem and use it as a basis for open exploration. You will apply scenario discovery techniques to identify key drivers of low reliability and explore structural explanations for the model behavior.
3. Gain hands-on experience in performing global sensitivity analysis on the predator-prey model, comparing the strengths and weaknesses of linear regression, Sobol, and extra-trees feature scoring methods.
4. Apply the best-performing sensitivity analysis methods (Sobol and extra-trees) to the lake problem. You will compare the influence of input parameters on the reliability objective and visualize feature importances for different release policies.
5. Develop a better understanding of complex systems and uncertainties, enabling you to prioritize efforts to manage uncertainties effectively and make well-informed decisions.

By the end of Weeks 3-4, you will have a solid foundation in scenario discovery and sensitivity analysis, deepening your understanding of model behavior under different uncertainties and policies. Armed with these skills, you will be better equipped to analyze complex systems and make more informed decisions in the face of uncertainty.

Suggested order of assignments

4. Scenario discovery - replication assignment: assignment 4 - scenario discovery.ipynb
   Replicate the results from influential papers in decision-making under uncertainty, applying the Patient Rule Induction Method (PRIM) to identify policy vulnerabilities, develop Shared Socio-economic Pathways (SSPs), and design adaptive decision-making strategies.
5. Lake model continued - subspace partitioning: assignment 5 - subspace partitioning lake model.ipynb
   Apply scenario discovery techniques to the lake problem, identify key drivers of low reliability, and explore structural explanations for the model behavior.
6. Sensitivity analysis - predator-prey model: assignment 6 - sensitivity analysis.ipynb
   Perform global sensitivity analysis on the predator-prey model, comparing linear regression, Sobol, and extra-trees feature scoring methods.
7. Sensitivity analysis - lake model: assignment 7 - sensitivity analysis lake model.ipynb
   Apply the best-performing sensitivity analysis methods (Sobol and extra-trees) to the lake problem, comparing the influence of input parameters on the reliability objective and visualizing feature importances for different release policies.