Model Calibration, Uncertainty & Global Sensitivity — Hands-on

Learn how to turn qualitative fits into quantitatively calibrated and stress tested systems models. This module walks through parameter estimation workflows, uncertainty quantification and global sensitivity analysis so that you can defend model predictions and focus on parameters and mechanisms that truly matter.

Model Calibration, Uncertainty & Global Sensitivity

Help Desk · WhatsApp

Session Index

Session 1 — Parameter Estimation & Calibration Basics Session 2 — Bayesian Calibration & Identifiability Session 3 — Uncertainty Quantification & Global Sensitivity Session 4 — Mini Capstone: Calibrated & Stress Tested Model

Session 1

Fee: Rs 8800

Parameter Estimation & Calibration Basics

Error models and objective functions for calibration

least squares and likelihood views measurement noise assumptions cost surfaces and minima intuition

Local optimisation and gradient based methods

Levenberg Marquardt style ideas handling bounds and constraints practical convergence diagnostics

Good practices for calibration experiments and data use

separation of training and validation data scaling and normalisation of readouts simple residual and fit checks

Session 2

Fee: Rs 11800

Bayesian Calibration & Identifiability

Bayesian view of parameter estimation

priors likelihood and posteriors credible intervals for parameters prediction intervals vs fits

MCMC and sampling concepts for calibration

random walk and adaptive schemes burn in mixing and diagnostics posterior summaries and plots

Structural and practical identifiability basics

non identifiable parameter combinations profile likelihood style thinking design hints to improve identifiability

Session 3

Fee: Rs 14800

Uncertainty Quantification & Global Sensitivity

Sources and types of uncertainty in systems models

parameter and structural uncertainty observation and process noise scenario and input uncertainty

Sampling based uncertainty propagation

Latin hypercube and Monte Carlo ideas prediction bands and fan plots communicating uncertainty clearly

Global sensitivity: screening and variance based methods

Morris screening concepts Sobol indices intuition linking sensitivity to design choices

Session 4

Fee: Rs 18800

Mini Capstone: Calibrated & Stress Tested Model

Select a small systems model and data set for calibration

Theory + Practical

Calibrate parameters and map uncertainty and sensitivity

run basic optimisation or sampling generate prediction intervals compute simple global sensitivity indices

Deliverables: calibrated model, UQ plots and sensitivity summary

SBML or notebook implementation tables of parameter and sensitivity metrics short interpretation and limitations note

PDF

Model Calibration, Uncertainty & Global Sensitivity Training | Parameter Estimation & UQ

Model Calibration, Uncertainty & Global Sensitivity — Hands-on