Lecture 3 Flashcards
What is the primary goal of sensitivity analysis?
What is the process to check the sensitivity of parameters in a model?
How can you determine which parameter is most sensitive in a model?
S(x) measures the sensitivity of the variable x to the parameter P
What happens on this slide regarding sensitivity?
Why are at least 3 observations per parameter required for calibration?
What is overfitting in the context of modeling?
How does sensitivity analysis differ when initial conditions are unknown?
What is the difference between Local Sensitivity Analysis (Local SA) and Global Sensitivity Analysis (Global SA)?
What is the advantage of Global Sensitivity Analysis (Global SA)?
What is Latin Hypercube Sampling, and how does it relate to Global SA?
What is the main difference between Latin Hypercube Sampling and random sampling?
How does Global SA handle correlation between parameters?
Local vs Global SA
Global SA
What is the purpose of sensitivity analysis?
To identify parameters that most influence model predictions.
What is local sensitivity analysis?
An analysis that varies one parameter at a time to observe its effect on model output.
What is global sensitivity analysis?
An analysis that varies multiple parameters simultaneously to understand their combined effects.
What is Latin Hypercube Sampling (LHS)?
A method that divides the parameter space into intervals to ensure comprehensive parameter coverage.
What is random sampling in sensitivity analysis?
A method that selects parameters randomly but may miss key interactions between variables.
What is a condition number (CN)?
A measure of the sensitivity of state variables to parameter changes. High CN values indicate significant impact, while low CN values suggest minimal influence.
What is overfitting in ecological models?
When a model fits noise rather than trends, leading to poor generalizability.
What is the recommended number of observations per parameter to avoid overfitting?
Between 3 and 10 observations per parameter.
