sensitivity-package: Sensitivity Analysis

In sensitivity: Sensitivity Analysis

Description

Methods and functions for global sensitivity analysis.

Details

The sensitivity package implements some global sensitivity analysis methods:

• Linear regression coefficients: SRC and SRRC (src), PCC and PRCC (pcc).

• Morris's "OAT" elementary effects screening method (morris).

• Bettonvil's sequential bifurcations (sb).

• Monte Carlo estimation of Sobol' indices: Sobol's scheme (1993) to compute the indices given by the variance decomposition up to a specified order (sobol), and Saltelli's scheme (2002) to compute first order and total indices with a reduced cost (sobol2002).

• Estimation of the Sobol' first order and total indices with Saltelli's so-called "extended-FAST" method (fast99).

Moreover, some utilities are provided: standard test-cases (testmodels) and template file generation (template.replace).

Model managing

The sensitivity package works either on R models than on external models (such as executables).

R models must be functions or objects that have a predict method, such as lm objects. Models defined as functions will be called once with an expression of the form y <- f(X) where X is the design of experiments, i.e. a data.frame with p columns (the input factors) and n lines (each, an experiment), and y is the vector of length n of the model responses (we say that such functions are vectorized).

If the model is external to R, for instance a computational code, it must be analyzed with the decoupled approach, see decoupling. This approach can also be used on R models that doesn't fit the specifications.

References

A. Saltelli, K. Chan and E. M. Scott eds, 2000, Sensitivity Analysis, Wiley.

sensitivity documentation built on May 2, 2019, 5:56 p.m.