In factor fixing, the model parameters that have the least influence on the model output are identified – often with the purpose of simplifying the model.
A total-effect sensitivity index quantifies the effect of the associated model parameter and all its interactions with other parameters on the model output. The total-effect sensitivity indices are well-suited for factor fixing.
Based on samples from a Monte Carlo simulation, the first-order sensitivity indices can be estimated without additional model calls. This requires the numerical approximation of a one-dimensional conditional expectation.
In factor prioritization, the input parameters are ranked according to their importance on the model output. The first-order sensitivity indices are a commonly applied sensitivity measure for that purpose.
A first-order sensitivity index quantifies the effect of varying the associated model input parameter alone. The first-order sensitivity indices are well-suited for factor prioritization.
We can decompose the variance of the model output in terms of a series expansion of combined effects if the model input parameters are statistically independent.
Derivative-based SA provides a local sensitivity measure that is commonly used to assess deterministic model parameters. By normalizing the derivatives with standard deviations, the measure can also be applied to uncertain parameters (of linear models).
Local sensitivity analysis assesses the influence of the input on the output around a specific point. Global sensitivity analysis assesses the influence of the uncertainty about the input on the output over the entire range of the input parameters.
Sensitivity analysis assesses how the input parameters and their uncertainties influence the model output. It is essential to define the goal of the analysis as the first step of any sensitivity analysis.
