Professional Certificate in Model Uncertainty Quantification Methods

This course is designed for data scientists, statisticians, and engineers seeking to enhance their skills in quantifying uncertainties in predictive models. Participants will gain proficiency in using advanced statistical methods and machine learning techniques to assess and manage prediction errors and model uncertainties effectively.

Students will learn to apply state-of-the-art uncertainty quantification techniques, including Bayesian methods, Monte Carlo simulations, and ensemble modeling. They will also develop skills in using software tools and programming languages such as Python and R for implementing these methods in real-world applications.

1. 1. Introduction to Uncertainty Quantification: Learners will study the basic principles and importance of uncertainty quantification in modeling, understanding how uncertainties arise and their impact on model predictions. They will gain foundational skills in identifying, characterizing, and representing uncertainties in models.
2. 2. Probability Theory and Statistical Inference: This module covers essential probability theory and statistical inference techniques necessary for quantifying uncertainties. Learners will learn to apply probability distributions, perform hypothesis testing, and use statistical methods to estimate uncertainties in model parameters.
3. 3. Sensitivity Analysis Techniques: Learners will explore various sensitivity analysis methods to identify which input variables significantly influence model outputs. Skills in conducting local and global sensitivity analyses, and interpreting the results for model improvement will be developed.
4. 4. Monte Carlo Methods: This module focuses on Monte Carlo simulation techniques for uncertainty quantification. Learners will gain expertise in generating random samples, simulating model behavior under uncertainty, and using Monte Carlo methods to compute statistics and propagate uncertainties.
5. 5. Bayesian Methods for Uncertainty Quantification: Here, learners will study Bayesian inference and its application in uncertainty quantification. They will learn to update prior knowledge with new data, perform Bayesian model calibration, and use Markov Chain Monte Carlo (MCMC) methods for parameter estimation.
6. 6. Uncertainty Propagation Techniques: This module covers methods for propagating uncertainties through models, including analytical and numerical techniques. Learners will understand and apply techniques such as polynomial chaos expansions and stochastic collocation methods.
7. 7. Model Calibration and Validation: Learners will learn advanced techniques for calibrating and validating models under uncertainty. This includes understanding the principles of model validation, developing strategies for model selection, and assessing model reliability and accuracy.
8. 8. Uncertainty Treatment in Decision Making: This module focuses on the application of uncertainty quantification in decision-making processes. Learners will learn to incorporate uncertainty into decision metrics, evaluate risk and uncertainty in decision-making, and use uncertainty quantification to inform decision strategies.
9. 9. Uncertainty Quantification in Complex Systems: Here, learners will explore the challenges and methods for uncertainty quantification in complex, multi-scale, and multi-physics systems. They will gain skills in handling high-dimensional problems, managing large datasets, and using advanced computational tools.
10. 10. Case Studies and Applications: This final module provides real-world applications of uncertainty quantification methods. Learners will analyze case studies from various fields such as engineering, environmental science, and finance, applying the skills and knowledge gained throughout the programme to solve practical problems.